Full-Text - International Journal of Computer Science Issues

Transcription

International Journal of
Computer Science Issues
Volume 3, August 2009
ISSN (Online): 1694-0784
ISSN (Printed): 1694-0814
© IJCSI PUBLICATION
www.IJCSI.org
IJCSI proceedings are currently indexed by:
© IJCSI PUBLICATION 2009
www.IJCSI.org
EDITORIAL
There are several journals available in the areas of Computer Science
having different policies. IJCSI is among the few of those who believe
giving free access to scientific results will help in advancing computer
science research and help the fellow scientist.
IJCSI pay particular care in ensuring wide dissemination of its authors’
works. Apart from being indexed in other databases (Google Scholar,
DOAJ,
CiteSeerX,
etc…),
IJCSI
makes
articles
available
to
be
downloaded for free to increase the chance of the latter to be cited.
Furthermore, unlike most journals, IJCSI send a printed copy of its issue
to the concerned authors free of charge irrespective of geographic
location.
IJCSI Editorial Board is pleased to present IJCSI Volume Three (IJCSI
Vol. 3, 2009). The paper acceptance rate for this issue is 37.5%; set after
all submitted papers have been received with important comments and
recommendations from our reviewers.
We sincerely hope you would find important ideas, concepts, techniques,
or results in this special issue.
As final words, PUBLISH, GET CITED and MAKE AN IMPACT.
IJCSI Editorial Board
August 2009
www.ijcsi.org
IJCSI Editorial Board 2009
Dr Tristan Vanrullen
Chief Editor
LPL, Laboratoire Parole et Langage - CNRS - Aix en Provence, France
LABRI, Laboratoire Bordelais de Recherche en Informatique - INRIA - Bordeaux, France
LEEE, Laboratoire d'Esthétique et Expérimentations de l'Espace - Université d'auvergne, France
Dr Constantino Malagón
Associate Professor
Nebrija University
Spain
Dr Mokhtar Beldjehem
Professor
Sainte-Anne University
Halifax, NS, Canada
Dr Pascal Chatonnay
Assistant Professor
Maître de Conférences
Université de Franche-Comté (University of French-County)
Laboratoire d'informatique de l'université de Franche-Comté, (Computer Science Laboratory of
University of French-County), France
Dr Deepak Laxmi Narasimha
Department of Software Engineering,
Faculty of Computer Science and Information Technology,
University of Malaya,
Kuala Lumpur, Malaysia
Prof N. Jaisankar
School of Computing Sciences,
VIT University
Vellore, Tamilnadu, India
IJCSI Reviewers Committee 2009
• Mr. Markus Schatten, University of Zagreb, Faculty of Organization and Informatics, Croatia
• Mr. Vassilis Papataxiarhis, Department of Informatics and Telecommunications, National and
Kapodistrian University of Athens, Panepistimiopolis, Ilissia, GR-15784, Athens, Greece,
Greece
• Dr Modestos Stavrakis, Univarsity of the Aegean, Greece
• Dr Fadi KHALIL, LAAS -- CNRS Laboratory, France
• Dr Dimitar Trajanov, Faculty of Electrical Engineering and Information technologies, ss. Cyril
and Methodius Univesity - Skopje, Macedonia
• Dr Jinping Yuan, College of Information System and Management,National Univ. of Defense
Tech., China
• Dr Alexis Lazanas, Ministry of Education, Greece
• Dr Stavroula Mougiakakou, University of Bern, ARTORG Center for Biomedical Engineering
Research, Switzerland
• Dr DE RUNZ, CReSTIC-SIC, IUT de Reims, University of Reims, France
• Mr. Pramodkumar P. Gupta, Dept of Bioinformatics, Dr D Y Patil University, India
• Dr Alireza Fereidunian, School of ECE, University of Tehran, Iran
• Mr. Fred Viezens, Otto-Von-Guericke-University Magdeburg, Germany
• Mr. J. Caleb Goodwin, University of Texas at Houston: Health Science Center, USA
• Dr. Richard G. Bush, Lawrence Technological University, United States
• Dr. Ola Osunkoya, Information Security Architect, USA
• Mr. Kotsokostas N.Antonios, TEI Piraeus, Hellas
• Prof Steven Totosy de Zepetnek, U of Halle-Wittenberg & Purdue U & National Sun Yat-sen
U, Germany, USA, Taiwan
• Mr. M Arif Siddiqui, Najran University, Saudi Arabia
• Ms. Ilknur Icke, The Graduate Center, City University of New York, USA
• Prof Miroslav Baca, Associated Professor/Faculty of Organization and Informatics/University
of Zagreb, Croatia
• Dr. Elvia Ruiz Beltrán, Instituto Tecnológico de Aguascalientes, Mexico
• Mr. Moustafa Banbouk, Engineer du Telecom, UAE
• Mr. Kevin P. Monaghan, Wayne State University, Detroit, Michigan, USA
• Ms. Moira Stephens, University of Sydney, Australia
• Ms. Maryam Feily, National Advanced IPv6 Centre of Excellence (NAV6) , Universiti Sains
Malaysia (USM), Malaysia
• Dr. Constantine YIALOURIS, Informatics Laboratory Agricultural University of Athens,
Greece
• Mrs. Angeles Abella, U. de Montreal, Canada
• Dr. Patrizio Arrigo, CNR ISMAC, italy
• Mr. Anirban Mukhopadhyay, B.P.Poddar Institute of Management & Technology, India
• Mr. Dinesh Kumar, DAV Institute of Engineering & Technology, India
• Mr. Jorge L. Hernandez-Ardieta, INDRA SISTEMAS / University Carlos III of Madrid, Spain
• Mr. AliReza Shahrestani, University of Malaya (UM), National Advanced IPv6 Centre of
Excellence (NAv6), Malaysia
• Mr. Blagoj Ristevski, Faculty of Administration and Information Systems Management Bitola, Republic of Macedonia
• Mr. Mauricio Egidio Cantão, Department of Computer Science / University of São Paulo,
Brazil
• Mr. Jules Ruis, Fractal Consultancy, The netherlands
• Mr. Mohammad Iftekhar Husain, University at Buffalo, USA
• Dr. Deepak Laxmi Narasimha, Department of Software Engineering, Faculty of Computer
Science and Information Technology, University of Malaya, Malaysia
• Dr. Paola Di Maio, DMEM University of Strathclyde, UK
• Dr. Bhanu Pratap Singh, Institute of Instrumentation Engineering, Kurukshetra University
Kurukshetra, India
• Mr. Sana Ullah, Inha University, South Korea
• Mr. Cornelis Pieter Pieters, Condast, The Netherlands
• Dr. Amogh Kavimandan, The MathWorks Inc., USA
• Dr. Zhinan Zhou, Samsung Telecommunications America, USA
• Mr. Alberto de Santos Sierra, Universidad Politécnica de Madrid, Spain
• Dr. Md. Atiqur Rahman Ahad, Department of Applied Physics, Electronics & Communication
Engineering (APECE), University of Dhaka, Bangladesh
• Dr. Charalampos Bratsas, Lab of Medical Informatics, Medical Faculty, Aristotle University,
Thessaloniki, Greece
• Ms. Alexia Dini Kounoudes, Cyprus University of Technology, Cyprus
• Mr. Anthony Gesase, University of Dar es salaam Computing Centre, Tanzania
• Dr. Jorge A. Ruiz-Vanoye, Universidad Juárez Autónoma de Tabasco, Mexico
• Dr. Alejandro Fuentes Penna, Universidad Popular Autónoma del Estado de Puebla, México
• Dr. Ocotlán Díaz-Parra, Universidad Juárez Autónoma de Tabasco, México
• Mrs. Nantia Iakovidou, Aristotle University of Thessaloniki, Greece
• Mr. Vinay Chopra, DAV Institute of Engineering & Technology, Jalandhar
• Ms. Carmen Lastres, Universidad Politécnica de Madrid - Centre for Smart Environments,
Spain
• Dr. Sanja Lazarova-Molnar, United Arab Emirates University, UAE
• Mr. Srikrishna Nudurumati, Imaging & Printing Group R&D Hub, Hewlett-Packard, India
• Dr. Olivier Nocent, CReSTIC/SIC, University of Reims, France
• Mr. Burak Cizmeci, Isik University, Turkey
• Dr. Carlos Jaime Barrios Hernandez, LIG (Laboratory Of Informatics of Grenoble), France
• Mr. Md. Rabiul Islam, Rajshahi university of Engineering & Technology (RUET), Bangladesh
• Dr. LAKHOUA Mohamed Najeh, ISSAT - Laboratory of Analysis and Control of Systems,
Tunisia
• Dr. Alessandro Lavacchi, Department of Chemistry - University of Firenze, Italy
• Mr. Mungwe, University of Oldenburg, Germany
• Mr. Somnath Tagore, Dr D Y Patil University, India
• Ms. Xueqin Wang, ATCS, USA
• Dr. Fondjo Fotou Franklin, Langston University, USA
• Mr. Haytham Mohtasseb, Department of Computing - University of Lincoln, United Kingdom
• Dr. Vishal Goyal, Department of Computer Science, Punjabi University, Patiala, India
• Mr. Thomas J. Clancy, ACM, United States
• Dr. Ahmed Nabih Zaki Rashed, Dr. in Electronic Engineering, Faculty of Electronic
Engineering, menouf 32951, Electronics and Electrical Communication Engineering
Department, Menoufia university, EGYPT, EGYPT
• Dr. Rushed Kanawati, LIPN, France
• Mr. Koteshwar Rao, K G Reddy College Of ENGG.&TECH,CHILKUR, RR DIST.,AP, India
• Mr. M. Nagesh Kumar, Department of Electronics and Communication, J.S.S. research
foundation, Mysore University, Mysore-6, India
• Mr. Saqib Saeed, University of Siegen, Germany
• Dr. Ibrahim Noha, Grenoble Informatics Laboratory, France
• Mr. Muhammad Yasir Qadri, University of Essex, UK
• Mr. Annadurai .P, KMCPGS, Lawspet, Pondicherry, India, (Aff. Pondicherry Univeristy,
India
• Mr. E Munivel , CEDTI (Govt. of India), India
• Dr. Chitra Ganesh Desai, University of Pune, India
• Mr. Syed, Analytical Services & Materials, Inc., USA
• Dr. Mashud Kabir, Department of Computer Science, University of Tuebingen, Germany
• Mrs. Payal N. Raj, Veer South Gujarat University, India
• Mrs. Priti Maheshwary, Maulana Azad National Institute of Technology, Bhopal, India
• Mr. Mahesh Goyani, S.P. University, India, India
• Mr. Vinay Verma, Defence Avionics Research Establishment, DRDO, India
• Dr. George A. Papakostas, Democritus University of Thrace, Greece
• Mr. Abhijit Sanjiv Kulkarni, DARE, DRDO, India
• Mr. Kavi Kumar Khedo, University of Mauritius, Mauritius
• Dr. B. Sivaselvan, Indian Institute of Information Technology, Design & Manufacturing,
Kancheepuram, IIT Madras Campus, India
• Dr. Partha Pratim Bhattacharya, Greater Kolkata College of Engineering and Management,
West Bengal University of Technology, India
• Mr. Manish Maheshwari, Makhanlal C University of Journalism & Communication, India
• Dr. Siddhartha Kumar Khaitan, Iowa State University, USA
• Dr. Mandhapati Raju, General Motors Inc, USA
• Dr. M.Iqbal Saripan, Universiti Putra Malaysia, Malaysia
• Mr. Ahmad Shukri Mohd Noor, University Malaysia Terengganu, Malaysia
• Mr. Selvakuberan K, TATA Consultancy Services, India
• Dr. Smita Rajpal, Institute of Technology and Management, Gurgaon, India
• Mr. Rakesh Kachroo, Tata Consultancy Services, India
• Mr. Raman Kumar, National Institute of Technology, Jalandhar, Punjab., India
• Mr. Nitesh Sureja, S.P.University, India
• Dr. M. Emre Celebi, Louisiana State University, Shreveport, USA
• Dr. Aung Kyaw Oo, Defence Services Academy, Myanmar
• Mr. Sanjay P. Patel, Sankalchand Patel College of Engineering, Visnagar, Gujarat, India
• Dr. Pascal Fallavollita, Queens University, Canada
• Mr. Jitendra Agrawal, Rajiv Gandhi Technological University, Bhopal, MP, India
• Mr. Ismael Rafael Ponce Medellín, Cenidet (Centro Nacional de Investigación y Desarrollo
Tecnológico), Mexico
• Mr. Supheakmungkol SARIN, Waseda University, Japan
• Mr. Shoukat Ullah, Govt. Post Graduate College Bannu, Pakistan
• Dr. Vivian Augustine, Telecom Zimbabwe, Zimbabwe
• Mrs. Mutalli Vatila, Offshore Business Philipines, Philipines
• Dr. Emanuele Goldoni, University of Pavia, Dept. of Electronics, TLC & Networking Lab,
Italy
• Mr. Pankaj Kumar, SAMA, India
• Dr. Himanshu Aggarwal, Punjabi University,Patiala, India
• Dr. Vauvert Guillaume, Europages, France
• Prof Yee Ming Chen, Department of Industrial Engineering and Management, Yuan Ze
University, Taiwan
• Dr. Constantino Malagón, Nebrija University, Spain
• Prof Kanwalvir Singh Dhindsa, B.B.S.B.Engg.College, Fatehgarh Sahib (Punjab), India
• Mr. Angkoon Phinyomark, Prince of Singkla University, Thailand
• Ms. Nital H. Mistry, Veer Narmad South Gujarat University, Surat, India
• Dr. M.R.Sumalatha, Anna University, India
• Mr. Somesh Kumar Dewangan, Disha Institute of Management and Technology, India
• Mr. Raman Maini, Punjabi University, Patiala(Punjab)-147002, India
• Dr. Abdelkader Outtagarts, Alcatel-Lucent Bell-Labs, France
• Prof Dr. Abdul Wahid, AKG Engg. College, Ghaziabad, India
• Mr. Prabu Mohandas, Anna University/Adhiyamaan College of Engineering, india
• Dr. Manish Kumar Jindal, Panjab University Regional Centre, Muktsar, India
• Prof Mydhili K Nair, M S Ramaiah Institute of Technnology, Bangalore, India
• Dr. C. Suresh Gnana Dhas, VelTech MultiTech Dr.Rangarajan Dr.Sagunthala Engineering
College,Chennai,Tamilnadu, India
• Prof Akash Rajak, Krishna Institute of Engineering and Technology, Ghaziabad, India
• Dr. Vu Thanh Nguyen, University of Information Technology HoChiMinh City, VietNam
TABLE OF CONTENTS
1. Pharmaco-Cybernetics as an Interactive Component of Pharma-Culture: Empowering
Drug Knowledge through User-, Experience- and Activity-Centered Designs – (pg 1-13)
Kevin Yi-Lwern YAP, Xuejin CHUANG, Alvin Jun Ming LEE, Alvin Jun Ming LEE, Raemarie
Zejin LEE, Lijuan LIM, Jeanette Jiahui LIM, Ranasinghe NIMESHA, NM5206 Project Team,
Communications and New Media Programme, Faculty of Arts & Social Sciences, National
University of Singapore
2. Similarity Matching Techniques For Fault Diagnosis In Automotive Infotainment
Electronics – (pg 14-19)
Mashud Kabir, Department of Computer Science, University of Tuebingen, D-72027 Tuebingen,
Germany
3 . Prototype System for Retrieval of Remote Sensing Images based on Color Moment and
Gray Level Co-Occurrence Matrix – (pg 20-23)
Priti Maheshwary and Namita Sricastava, Deparment of Mathematics, Maulana Azad National
Institute of Technology, Bhopal, Madhya Pradesh, India
4. Performing Hybrid Recommendation in Intermodal Transportation – the FTMarket
System’s Recommendation Module – (pg 24-34)
Alexis Lazanas, Industrial Management and Information Systems Lab, University of Patras, Rion
Patras, 26500, Greece
5. Geometric and Signal Strength Dilution of Precision (DoP) Wi-Fi – (pg 35-44)
Soumaya Zirari, Philippe Canalda and François Spies, Computer Science Laboratory of the
University of Franche-Comté, France
6. Implementation of Rule Based Algorithm for Sandhi-Vicheda Of Compound Hindi
Words – (pg 45-49)
Priyanka Gupta and Vishal Goyal, Department of Computer Science, Punjabi University Patiala
IJCSI International Journal of Computer Science Issues, Vol. 3, 2009
ISSN (Print): 1694-0814
1
Pharmaco-Cybernetics as an Interactive Component of Pharma-Culture: Empowering Drug
Knowledge through User-, Experience- and Activity-Centered Designs
Kevin Yi-Lwern YAP1,2, Xuejin CHUANG2, Alvin Jun Ming LEE2, Raemarie Zejin LEE2, Lijuan LIM2, Jeanette Jiahui LIM2
and Ranasinghe NIMESHA2
1
2
Department of Pharmacy, Faculty of Science, National University of Singapore
Block S4, 18 Science Drive 4, Singapore 117543, Singapore
NM5206 Project Team, Communications and New Media Programme, Faculty of Arts & Social Sciences, National University
of Singapore
Abstract
The advent of the World Wide Web (WWW) has led to the
creation of many web publishing platforms. Patients are
becoming more well-informed through drug and health-related
information over the internet. The integration of interactive
media technologies and the WWW provides an opportunity to
improve the pharmaceutical care of patients on anticoagulant
therapy. In this paper, the concept of ‘pharmaco-cybernetics’ is
introduced through the creation of an interactive tool which
consists of a pill-catching game and hangman game designed to
enable users to learn about warfarin tablet strengths and drug
interactions, based on user-centered (UCD), experience-centered
(ECD), and activity-centered design (ACD) approaches.
Currently, this tool is largely based on UCD and ECD. However,
the potential of incorporating the ACD approach in the tool’s
design is definitely attractive. Pharmaco-cybernetics can
empower patients with the appropriate knowledge regarding their
therapy so that they can better participate in the management of
their health.
Key words: Drug Information, Interactive Games, PharmacoCybernetics, User Interaction, Warfarin.
1. Introduction
Anticoagulation therapy involves the use of drugs to
help prevent and treat blood clots in the arteries or veins.
Anticoagulants, also known as ‘blood thinners’, work in
various ways to inhibit blood-clotting factors in the body.
Warfarin is an oral anticoagulant which works by blocking
the action of vitamin K in the liver. It is usually prescribed
for people with certain types of cardiovascular conditions
or those suffering from deep vein thrombosis [1]. Patients
on warfarin therapy are usually treated for a period of time
ranging from a few months to long term chronic therapy.
The dose of warfarin taken by the patient is adjusted
according to the results of a blood test known as the
International Normalized Ratio (INR), which is a measure
of how long a patient’s blood takes to clot. An INR above
or below a set target means that the patient is at a higher
risk of bleeding and clotting occurrences respectively.
Thus, the dose of warfarin has to be individualized
according to the patient’s response to the drug.
Warfarin comes in many brands. Patients are advised
not to switch among brands as different brands have
slightly different efficacy. In Singapore, the brand
Marevan® is used, and it comes in a tablet with three
strengths which can be identified by its color: 1mg
(brown), 3mg (blue) and 5mg (pink). Patients on warfarin
therapy may need adjustment of their dosages until their
INR stabilizes, and this may be confusing for some
patients, especially during the initial stages. Hence, it is
important to educate them to recognize the tablets which
they are taking and remember the dosages of their therapy.
It is easier for the patient to remember the dosage if they
can correlate it with the strength of the tablets, which in
turn, can be identified by their colors.
Warfarin also has many drug interactions. In a broad
sense of this paper, these include other medicines,
nutritional supplements, traditional herbs, and foods which
are rich in vitamin K. It is prudent that patients on
warfarin therapy also know some of its common
interactions so that they can adapt to any changes in their
dietary habits and lifestyles.
In traditional medical practice, healthcare professionals
have always played active roles in the care of patients. For
example, doctors tell their patients what is wrong and how
to get better, and pharmacists counsel patients with
regards to their medications. For warfarin therapy, patients
currently see a pharmacist-run clinic for counseling, where
they are educated about the drug itself and how to
recognize and manage signs and symptoms of adverse
effects and drug interactions. In addition, they are also
given supplementary materials such as pamphlets as part
of their education. However, the patients’ understanding
of warfarin therapy is limited to the time for each
counseling session, and the frequency in which they re-
2
visit the clinic for follow-up. Thus, their knowledge on
warfarin may be limited, particularly for those who are on
this medication for the first time. The lack of knowledge
or misinterpretation of information about the drug or its
use can affect their compliance to their medication, which
may consequently lead to the patients suffering from drugrelated problems (DRPs) such as under- or overdosing, or
potential drug-drug, drug-food or drug-herb interactions
[2]
.
Human-computer interaction (HCI) has become a norm
in society. The roles between patients and healthcare
professionals have evolved with the information age.
Internet and informatics technologies brought about by the
cyber era have been critical in transforming the public’s
attitudes towards healthcare and medicine. The interface
between HCI and health services has led to the birth of
medical informatics, which aims to develop studies and
instruments to solve clinical issues in the practical setting
[3]
. Its ultimate goal is to improve the healthcare of
patients. As such, many issues from the genetics, social,
economic and environmental factors, cognitive, emotional
and behavioral domains can also play a role [4]. The
emergence of the World Wide Web (WWW) is one of the
most significant developments in the history of the internet
[5]
. The internet is rapidly gaining importance not just for
healthcare professionals, but for patients as well. Although
healthcare professionals access information on the internet
to help them make decisions regarding patient care,
patients are also becoming more well-informed about their
health and health-related issues through the information
which they can get over the internet. Patients are now just
as likely to be able to highlight the risks, various therapies
and available treatments to their healthcare providers [6].
As traditional therapy is being translated to the internet,
the layman is now more aware of his health and is able to
better understand the science behind the various illnesses
through information he gets from the WWW. Albeit the
uncertainty as to whether cybermedicine will ever be
comparable to non-cybermedicine [7], the WWW has
nevertheless impacted the way healthcare is being
practiced today. The challenge is for both healthcare
professionals and patients to critically evaluate the vast
amounts of available information so as to provide the best
care for the patients’ well-being.
1.1 The Roles of the Internet and Interactive
Media in Healthcare
The traditional role of media in healthcare has involved
the use of audio and video programs in public health
education, such as with psychiatric diseases, cancer and
smoking. Film and photography were used as forms of
‘Edutainment’ – an Education-Entertainment strategy – to
address the stigma of people experiencing depression [8]
and schizophrenia [9]; while the American Cancer Society
leveraged the use of movies as an educational tool for the
public on cancer in the 1920s [10]. In fact, popular
Hollywood films in the 1930s to 1970s also used this
strategy to portray some cancers as being more ‘favorable’
since they were more photogenic and less offensive [11].
Furthermore, a recent trial also showed the usefulness of
digital media in improving the knowledge and awareness
of prostate cancer screening among African-American
men [12]. However, the two most pressing health-related
issues currently which involve the impact of digital media
are on its effects on the views and attitudes of sexuality [13]
and smoking among youths [14,15].
In recent years, the internet has become a very popular
HCI tool in a person’s daily life. It is not uncommon
nowadays for patients to search for health-related
information online. The World Wide Web Consortium
(W3C) [16] and the Internet Engineering Task Force (IETF)
[17]
have not only provided common standards for data,
information and software applications for the WWW, but
also encouraged users to discuss about various internetrelated operational and technical problems. Users can now
navigate through a vast and complex web of linked
computer documents through an inexpensive, easy-to-use,
cross-platform, graphic interface which supports items like
buttons, scroll lists, tables and pop-up menus for user
interaction. However, the current hype in healthcare not
only embarks on the use of IT and the WWW, but also the
integration of interactive media technologies. Interactive
media not only establishes a two-way communication
among its users, but allows active participation as well. An
opportunity exists for web users to gain information and
knowledge in a more interesting manner. Internet
interactivity can exist in both digital and multimedia forms,
and is most commonly represented by means of text, audio,
video, graphics, images and animation [18]. As long as one
has the hardware, software, talent and skills for
developing an interactive application, it can be mounted
on the WWW through inexpensive browsers.
1.2 Animation
Healthcare
as
an
Interactive
Tool
in
Animations have always been promoted as a way to
showcase the dynamics of user interface actions. People
encounter animations frequently since they have been used
for various purposes, particularly in web pages and online
advertisements. Animations are useful for presenting
highly abstract or dynamic processes, or when the user is
involved in an action or process [19]. It is known that user
satisfaction with animations is usually quite high, unless
they distract the user from focusing on key issues [20]. The
applications of animation are widespread, normally
involving the entertainment and advertising industries.
However, this form of interactivity is also getting more
widely accepted in the healthcare world.
There are many examples of animation applications in
the medical sciences, such as in medicine and dentistry [21],
orthopedics [22,23], and aesthetics surgery [24,25]. A virtual
human simulation using a 3D phantom was developed by
Oak Ridge National Laboratory and its collaborators [26] at
the beginning of the century as a computer representation
of the human anatomy. Animated films can also be used in
the field of psychology for teaching purposes, such as
characterizing personality types. An example can be
extracted from the animated film ‘Who Framed Roger
Rabbit’ [27], in which Roger exhibits a whole range of
personality traits from being extroverted and aggressive to
being insecure and anxious. However, film animation is
only one of animation techniques that can be used in the
health sciences.
Advancements in computer technology have
revolutionized the way healthcare is practiced. As
computers become more affordable and newer
technologies emerge, traditional animation techniques of
tweening and morphing have transformed into
computerized versions created by two- (2D) and threedimensional (3D) bitmap and vector graphics. The
development of the WWW has led to the creation of many
web publishing platforms, including HyperText Markup
Language (HTML) and its variants, Java applets, Flash
and Shockwave, among others. Web technologies have
also enabled the generation of other forms of web pages
like Hypertext Preprocessor (PHP) and Active Server
Pages (ASP). HTML has been the well-known standard
format for publishing content on the WWW, but its
limitation lies in the management of interactive and
animated content. However, the WWW has now managed
to successfully integrate Flash technology for this purpose
due to its advantages of not having cross-platform and
cross-browser compatibility problems, and the ‘Flash
everywhere’ phenomenon is getting very popular with
website developers [28]. Websites can now be created using
a combination of HTML and Flash, or created entirely in
Flash. A recent small-scale usability study done by
Piyasirivej reported that users generally enjoy Flash sites
more than HTML sites [28]. Examples are the ‘Virtual Knee
Surgery’ and ‘Choose the Prosthetic’ games developed by
Edheads + COSI where the user takes on the role of a
virtual surgeon to diagnose knee replacement patients and
carry out a total knee replacement surgery [29]. However,
despite the attractiveness of such technologies in the
various areas of healthcare, their progress in the
pharmaceutical arena is still slow.
3
1.3 Pharmaco-Cybernetics as Part of PharmaCulture
The objectives, roles and value-addedness of clinical
pharmacists have always been in continuous debate.
Nevertheless, many organizations such as the World
Health Organization (WHO) and the Nuffield Foundation
have recognized pharmacists as essential health care
providers [30]. The practice of pharmaceutical care forms
the cornerstone of clinical pharmacy, and its concept
revolves around identifying, solving and preventing drugrelated problems (DRPs) with regards to a patient’s drug
therapy [31]. Although this area has significantly
contributed to new approaches in pharmacy education,
several ‘driving forces’ that will impact the value of
pharmacists have been identified [30]. These include: (a)
improved care and protection for patients, especially the
chronically ill or those with particular types of diseases
(e.g. acquired immune deficiency syndrome or AIDS); (b)
training new pharmacy professionals to be more patient
orientated; and (c) the need for advanced pharmaceutical
expertise and new skills to keep up with accelerated
information technology so as to be able to manage new
treatments.
Pharmaco-cybernetics is an upcoming area of pharmacy
which involves advanced skills and expertise to deal with
HCI concepts and technologies in relation to medicines
and drugs. The term ‘pharmaco’ is derived from the Greek
term ‘pharmakon’ meaning drugs or poisons [32], and
‘cybernetics’ comes from the Greek term ‘kubernetes’,
which can be translated to mean ‘the art of steering’ [33,34].
Originally defined by Norbert Wiener in his book of the
same title, he defined ‘cybernetics’ as the science or study
of ‘control and communication in the animal and the
machine’ [33-35]. Aptly described by the American Society
for Cybernetics (ASC) as the design, discovery and
application of principles of regulation and communication
[35]
, this is a multi-disciplinary area which has been applied
to many fields such as system theory, psychology,
anthropology, sociology, and more recently, biology,
engineering and computer science [34]. The single
characteristic that defines a cybernetic system is the
relationship between endogenous goals and the external
environment [36]. In fact, this was not a new concept in
healthcare, and was already applied in the 1970s by Maltz
as a means of setting goals of positive outcomes for his
patients who were not satisfied by their plastic surgery
procedures [37]. However, the traditional concept of
cybernetics has evolved into a modern theory known as
‘new cybernetics’ or ‘second-order cybernetics’, in which
information is viewed as construct and reconstructed by
individuals interacting with the environment [38,39]. This
means that the system is not only dependent on the
4
observer or person interacting with it, but it also links the
individual with the society as a whole.
The science of cybernetics has further led to the term
‘cyberspace’ being coined by Gibson in his famous book
Neuromancer, which identified a virtual representation of
information in varying states of accessibility, linked to
various people and organizations [40-42]. A similar concept
was brought up in the movie ‘The Matrix’ and its sequels
in which Neo, a computer programmer, who lived in a
future world perceived by humans as reality, was actually
a simulated matrix created by sentient machines to subdue
the human race [43]. This term is now ubiquitously used to
describe anything which is associated with computers,
information technology, and the internet. It also
incorporates the elements of social experiences and
interaction of individuals through the exchange of ideas
and the sharing of information [44].
Thus, ‘pharmaco-cybernetics’ or ‘pharma-cybernetics’
aptly describes the science of dealing with medicines or
drugs through applications of HCI concepts and
technologies so as to reduce or prevent DRPs, and
ultimately, improve pharmaceutical care in patients. It
involves communication and feedback with the users, and
connects control (i.e. actions taken in the hope of
achieving goals) with communication (i.e. the flow of drug
information and knowledge between the user and the
cybernetic system or environment).
In this paper, we attempt to introduce the concept of
‘pharmaco-cybernetics’ through the creation of a simple
interactive tool aimed at improving the knowledge of users
on anticoagulation therapy. In particular, two prototype
games which are targeted at students in the pharmaceutical
sciences and patients on warfarin therapy will be
discussed. Ten web animation principles [45], as well as
user- (UCD), experience- (ECD) and activity-centered
design (ACD) approaches which can be considered in the
designing of pharmaco-cybenetic systems will also be
elaborated through a critique of the tool based on a pilot
usability survey that was done. Due to space constraints,
only important concepts related to the design frameworks
will be discussed. The reader is referred to Appendices 1,
2 and 3 for more detailed application summaries.
2. Creation and Evaluation of WarfarINT
The WarfarINT interactive tool was created as an
information resource for patients, students and the general
public who are interested in learning about anticoagulation
therapy. WarfarINT stands for ‘Warfarin INTerative’, and
consists of 2 games (Fig. 1) which provides the interactive
component for users.
The first is a pill-catching game in which users have to
catch different colored warfarin tablets dropping from the
top of the screen by moving a pill bag with their mouse in
a horizontal direction. Their scores are correlated with the
strength of the tablets that are caught, which in turn are
reflected by the different colors. The second is a hangman
game in which users are supposed to guess a drug, food or
herb that interacts with warfarin. The objectives of this
tool are to enable users to correlate the tablet colors with
their strengths, as well as know the drugs, herbs or foods
that interact with warfarin in an interesting manner.
Fig. 1 Screenshots of the interaction tool which consists of 2 games: (a)
Warfarin Game, and (b) Warfarin Hangman.
A pilot usability study was also carried out on a group
of pharmaceutical science students at a local educational
institution to evaluate how well the interactive tool helped
in improving their knowledge of the anticoagulant drug.
Participants were given 15 minutes to answer a
questionnaire which consisted of questions categorized
into 3 parts: (a) user demographics, (b) general knowledge
and views on anticoagulation therapy and online
interaction tools, and (c) feedback and experiences on
using the interactive tool (warfarin games). A fifth of the
time (3 minutes) was dedicated to playing the games. The
results were then evaluated based on descriptive statistics
and participants’ responses.
A total of 25 participants were recruited in the study,
with a response rate of 92%. Two responses were
excluded from analysis due to incomplete submissions.
The mean age of the respondents was 19.7+/-0.8 years,
and majority were females (87%). All respondents had
previously heard of warfarin before participating in the
study, but did not know about its tablet strengths and
interactions.
5
3. Human-Computer Interaction Frameworks
in Pharmaco-Cybernetics
3.1 The User-Centered Design (UCD) Approach
User-centered design (UCD) is a broad term used to
describe design processes in which end-users play a role in
influencing how a product’s design takes shape. Users are
placed at the center of the design process throughout the
planning, creation and development phases of the product.
The concepts of visibility, mapping and feedback play
crucial roles in the UCD approach [46].
The principle of visibility states that the user should be
able to figure out the use of a product based on the
visibility of its components. In other words, the product’s
parts or components should convey a correct message
regarding its usage [46]. This can be correlated to the
animation principles proposed by Weir and Heeps
(Appendix 1) [45].
The product, in this case is the tool consisting of the
games, should not distract users’ attention from salient
information, but rather, convey its intended message
across. Users should be drawn to the essential features of
the animation so that they can focus on the relevant
aspects. The graphical user interfaces (GUIs) of the tool
(Fig. 2) are located in the middle of the webpages so that
the user’s attention will be focused on the games. The
white backgrounds of the webpages are meant as contrasts
to the background of the games, and the titles of the games
are kept simple and self-explanatory so that first-time
users would know what to expect of the tool.
Fig. 2 Graphical user interfaces of the (a) warfarin pill-catching and (b)
hangman games.
In addition, visibility was demonstrated in the games
through short and concise instructions to users on what the
games entail and how to play:
“Collect as many warfarin tablets as you can! Move
your mouse to shift the pill bag left and right. Each tablet
color awards you points equivalent to its strength.” –
Instructions of the pill-catching game.
“Choose a letter by clicking on it… The letter changes
to green if your guess is correct, and red if your guess is
wrong.” – Instructions of the hangman game.
The use of ‘backup’ text to provide additional details can
help users understand the rationale of the animation better
provided it is used sparingly. Animations combined with
text and sound can reduce the likelihood of an ambiguity
in interpretation by the user. However, when used
inappropriately, it may cause distractions and cognitive
overloads.
Besides textual information, sounds can also support
ambiguity and provide feedback to the users regarding
certain results. However, it should only be used to enhance
the purpose of the animation. When used inappropriately,
sounds can confuse the user instead of enhancing their
information-retrieval experience. In the pill-catching game,
users would hear a ‘boing’ when they manage to catch a
tablet, but if they miss, a ‘splash’ would be heard instead.
This enables the users to discriminate between a score and
a miss, which would be important since the users would
6
strive to hear more ‘boings’ than ‘splashes’ to gain higher
scores.
The use of appropriate colors and adherence to color
conventions are also important for visibility of the product.
Like sounds, irrelevant color differences can also distract
and mislead users of the product. Colors are more than just
a cosmetic effect. They do not only help convey messages
to users, but also affect the users’ perceptions of depth and
space. The colors of the animated tablets follow the actual
color convention of warfarin tablets in reality with regards
to their tablet strengths. A 3D aspect is also achieved in
the hangman animation through the use of different colors.
A brown surface with red diagonal lines gives the ground
a horizontal effect, and the pole and stool seem to be
situated on the ground. The background is green to
distinguish it from the other objects in the animation, and
to give a sense of calm to the user playing the game, since
green is often associated with safety (e.g. traffic lights) or
nature (e.g. trees).
Humans have limited visual processing capability.
When faced with a visually cluttered display, users tend to
ignore some components in their perceptual field, and this
often impedes the delivery of the intended message. To
avoid clutter of our online tool, the animation screens are
centralized in the middle of the webpages (Fig. 2). In the
pill-catching game, the title, instructions, and scores, are
placed on the top left and right corners respectively. The
button to start and restart the game, indicated by ‘Play
Again’, is placed below the ‘Game Over’ message so that
users can click on it to play the game. Similarly, the title
and instructions of the hangman game occupy the top half
of the screen, and the animation of the hangman is located
just beside the words that users are supposed to guess, so
that they know how many wrong guesses they have made.
Clutter is also minimized as users are allowed to expand or
collapse the categories of drug interactions as appropriate.
Mapping [46], the second principle of UCD, describes the
link between one’s intended actions (what one wants to do)
to actual operations (what appears to be possible). In
animated products, it is crucial for the designer to
appreciate the insight of semiotics. Users will be able to
play the games if the games can be mapped to processes or
objects that are known or familiar to them. The target
audiences of the games are pharmacy/ pharmaceutical
science students and patients on warfarin therapy, who are
expected to be familiar with the drug. Furthermore, users
can guess the interactions based on their previous
experience of knowing how to play the hangman game.
Proper positioning and organization of objects in the
games can help users understand how to play the games.
The tool uses natural mapping of the left-right clicks on
the mouse controls that are familiar to users. This leads to
an immediate understanding of how to use these controls
to play the games. Incorporating these controls in the
games allows for easier manipulations of the various
animated components such as moving the pill-bag to catch
the dropping warfarin tablets, and selecting the alphabets
of the interacting drug. Gestalt’s law of proximity which
states that ‘related items should be placed closer together
than non-related items’ also applies here. Similarly,
information deemed to be of greater importance should
appear in positions of greater importance on the screen
from the user’s perspective. Related items in the games are
grouped together in time, space and shape, such as with
the warfarin tablets dropping in a vertical direction while
the pill bag moves in the opposite horizontal direction; and
the hangman animation being grouped side-by-side with
the word of the interacting drug. Users who play the
games will then be able to better remember the warfarin
interactions, as well as the tablet strengths.
For animations, the duration of exposure to users also
affects their ability to interpret and understand the
information about the product. Too short an exposure time
will leave the viewer confused, but too long a time can
lead to boredom and fatigue. Both games provide an
adequate amount of exposure time to users – the pillcatching game lasts less than a minute so that users do not
get bored, yet have enough time to learn and correlate the
tablets’ colors with their strengths; while users are given
an option to end the hangman game in the middle of
gameplay or if they give up guessing the word, or else,
frustration will result and lead to the user not wanting to
play the game again. Generally, if the correct amount of
information exposure cannot be determined, the common
rule of ‘too-much is better than too-little’ can be applied.
A principle that deserves special mention in this paper is
that of complying with the Co-operative Maxims. Based
originally on Grice’s Coorperative Principle, Weir and
Heeps have defined them with regards to animation in
terms of (a) quality (the animator tells/ portrays the truth),
(b) quantity (the intended message is adequately conveyed
without use of excess animation), (c) relation (the
animations are organized in a meaningful order), and (d)
manner (the animations are clear and natural, avoiding
ambiguity and disorder). The warfarin tool follows these
principles in the form of simple instructions and
information that is easily understood by the layman, with
the exception of drug names which cannot be simplified,
so as to avoid misinterpretation and ambiguity. Similarly,
these principles can and should be applied in any tool/
product that are designed for the purpose of providing
drug information. The explanations of these ‘Four
Pharmaco-cybernetic Maxims’ are provided in Table 1.
Table 1: The ‘Four Pharmaco-cybernetic Maxims’ for designing
pharmacy and/or pharmaceutical science tools.
Design
Explanation of principle with regards to pharmacy
principle
and/or pharmaceutical sciences
Quality
Drug information content provided by the informatics or
Quantity
Relation
Manner
internet tool(s) should be accurate and follow appropriate
resources for evidence-based therapies (e.g. research
articles, established databases or product information).
Adequate information about the drug or drug therapy is
provided so that users of the tool know enough to minimize
the likelihood of drug-related problems (e.g. underdose,
overdose, drug interactions).
Drug information provided by the tool(s) is/are relevant to
what the target audience needs to know, and should clarify
their doubts instead of making them more confused.
Drug information provided by the tool(s) is/are conveyed
clearly in an appropriate manner which avoids ambiguity
and misinterpretation (e.g. layman language for the patient
and medical jargon for healthcare professionals).
In UCD of products, feedback is largely a crucial
component as it reflects to the user about what action has
been done and what result is achieved [46]. Feedback is
accomplished in the warfarin tool as the user seeing the
pill bag move in response to his mouse movements, and
parts of the hangman animation or the letters appearing as
part of the word when he selects wrong or correct
alphabets respectively.
Feedback in animated tools should also follow the
traditional features developed by Walt Disney Studios,
which aims to make animations as realistic and
entertaining as possible. The ‘Squash and Stretch’ and
‘Timing and Motion’ aspects are most commonly accepted
by the public. The former defines an object’s rigidity and
mass by distorting its shape during an action, and the latter
follows the natural motion of an object such as
acceleration and deceleration, moving in curved paths, or
experiencing color and texture changes. Potentially
‘unreal’ aspects of an animated object’s behavior could
hinder users from interpreting the correct message.
‘Squash and Stretch’ in the games (Fig. 3) is demonstrated
by the distorting/ shrinking of the pill bag when the user
catches the tablet and the rope becoming taut when the
hangman is no longer supported by the stool. On the other
hand, ‘Timing and Motion’ is seen through the
acceleration of the dropping tablets and the hangman and
his feet dropping lower when the stool topples. These give
users the perceptions of gravity and friction in the
animations, which translates a sense of virtual reality when
playing the games.
7
Fig. 3 ‘Squash and Stretch’ aspect in the pill-catching game, and ‘Timing
and Motion’ aspect in the hangman game.
Users are a central part of the UCD developmental
process. Although UCD is about engineering usability, it
fails to take into account other important elements such as
environmental and socio-cultural factors. In the creation of
the interactive tool, it was assumed that all users would be
familiar with the mouse even though some users might be
more familiar and comfortable playing the games with the
keyboard instead. The games also did not take into
account the varying educational levels, or the settings
and/or situations in which potential users would be using
this online interactive tool. This is a condition known as
‘design myopia’ which is characterized by the shortsightedness of the designer. To the designer, the product
may appear suitable, even ideal. Yet, to the common
layman, the same product may seem unobvious and
obscure. This can result in an ‘adverse outcome’ of
breaking the user’s focus in the games and hindering his
learning potential. One approach to solving this problem is
to seek ‘fresh eyes’ on the product through means of usertesting to ensure that a suitable product is produced for the
intended purpose, and is also efficient and effective during
its development [47]. In this case, the pilot study was to
minimize possible misinterpretations and potential
problems before the product is released on a larger scale to
patients and pharmacy undergraduate students.
The results showed that although 75-85% of the
respondents deemed the instructions of the games to be
clear, one respondent actually commented to “Give some
instructions on playing the games” as a free-response
feedback. This situation could not have been predicted or
detected if a usability study had not been carried out on
the games. The participants in our pilot study had different
requirements and experiences with the games, and this
proved to be one of the major limitations of UCD which
can be accounted for by experience-centered (ECD) and
activity-centered designs (ACD), discussed in later
sections. Thus, there is a need to involve potential users in
the environment in which the interactive tool would be
used so as to increase its effectiveness, and consequently,
its acceptance and success.
8
3.2 The Experience-Centered Design (ECD)
Approach
Norman’s principles on emotional design stem from our
varied responses towards everyday things. The variables
that deliver a positive emotional experience vary greatly
with the appearance or functioning of a tool [48], and can
be matched with the visceral, behavioral and reflective
levels of design [49].
At the visceral level, the physical features of a product
(e.g. look, feel, sound) dominate over an otherwise usable
but plain looking product [49]. The current designs of the
warfarin tool are meant to pique the users’ interest in
playing the games. However, from our results, 10-20% of
the respondents rated the visual appeal as ‘fair’ even
though majority (45-70%) rated it ‘good’ to ‘excellent’.
This suggests that both games could be improved with
more aesthetically pleasing designs so as to give users a
thrill during gameplay which will enhance their overall
experience [49].
The behavioral level sees functionality as being
paramount [49]. The pill-catching game affords function
and usability through the user’s mouse movements as an
‘instinctive’ extension of his hand to move the pill bag to
catch the dropping tablets; while the hangman game does
this by leveraging on the user’s prior experience of
playing the ‘pen-and-paper’ version. Feedback is present
through real-time score updates in the pill-catching game,
and the various stages of hanging in the hangman game.
However, the underlying objectives of the games are not
explicitly made known to the user. Users may find it
difficult to keep track of their scores while simultaneously
trying to relate it to the strengths of the tablets. Similarly,
users who do not know any warfarin interactions would
not find the game useful. To further improve on the
behavioral aspects, immediate feedback on the scores and
the tablet strengths can be expressed through a storyline,
such as a better health-related outcome of a virtual patient,
and increasing the sizes and color intensities of the tablets
with higher strengths. Providing the interaction effects of
the drug, herb or food will also allow the user to
understand the need of knowing the drug interactions.
The reflective level [49] is related to the ‘emotional
thread of experience’ by McCarthy and Wright which
describes personal meaning derived from use of a product
[50]
. Sixty-five percent of the survey respondents thought
that the interactive tool did help them learn about warfarin,
even though it took a while for the learning to be
assimilated. The factors that could probably keep them
motivated in playing the games are the high scores in the
pill-catching game, since they indicate the user’s level of
accomplishment, and he is motivated to better his scores
and learn about the tablet strengths; and the congratulatory
message indicating “[the hangman] is alive!” when the
user guesses the word correctly. This gives meaning and
satisfaction to the user when he saves the hangman.
However, if he loses, words of encouragement “Don’t give
up!” appear to motivate him to play another round.
The ‘sensual thread’ describes the involvement of the
human senses in shaping an experience [50]. Both games
currently focus on sight and utilize the user’s experience
of moving and clicking the mouse to play. Sound effects
which provide feedback when the user catches (‘boing’) or
misses (‘splash’) a tablet cater to his sense of hearing.
However, the user plays the hangman game in silence.
Short midi, wav or mp3 files to indicate a win or loss in
the game can further enhance the user’s experience in this
case. Mounting the games on other platforms such as
personal digital assistants (PDAs) or iPhones can also
provide touch-alternatives and a completely different
experience to mouse-clicking.
The ‘compositional thread’ describes how one frames
the many parts that make up one’s whole experience [50].
According to this principle, the games should be
considered in relation to the rest of the WarfarINT website.
A common feedback from the survey was the lack of
adequate information about the drug. Although this could
be due to the limited time given in the pilot study to
explore the rest of the website, this was seen as a
‘breakdown’ by the respondents as the games seemed to
be relatively disjointed from the rest of the website.
Questions such as “how do these things go together” and
“I wonder what will happen if [action occurs]” could not
have been answered by the users. Thus, an improvement
would be to include the warfarin dosing information on
the same page as the pill-catching game instead of a
separate page, as is the current case. Another suggestion
from the respondents was to “show image[s] of the food
interaction with the correct word” in the hangman game
for a more positive and added visceral feel to the
experience.
The ‘spatio-temporal’ thread describes one entering a
state of ‘flow’ as he becomes engrossed in his experience
[50]
. Both games managed to keep the respondents
engrossed in gameplay, with 55% and 70% of the
respondents indicating that their levels of concentration
increased during continuous gameplay of the pill-catching
and hangman games respectively. However, some
comments from the respondents also suggested to “make
the pill catching game more interesting.” This can be done
by splitting the game into varying difficulty levels and an
animated storyline, for example, a virtual patient whose
blood vessels become less blocked due to the bloodthinning effect of warfarin, resulting in the patient
improving from his medical condition. On the other hand,
only users who have adequate drug vocabulary knowledge
of the warfarin interactions (e.g. pharmacy or medical
students) are immersed in a state of flow when playing the
hangman game. Patients who might not be as well-versed
in the interactions might suffer from a ‘disruption of flow’
due to frustration of not getting the correct word. Hints
can be provided in this case to ease the current steep
learning curve of the game.
The designing of interactive systems require an
understanding of how a person experiences the product
from an interaction-centered viewpoint [51]. Cognitive
user-product interactions require users to focus on the
product at hand, thus users of both games have to learn
what their actions will lead to during gameplay. It was
suggested in the survey that the warfarin tablets drop too
quickly in the pill-catching game, and that users could not
keep track on their scores without comprising their
gameplay. Increasing tablet sizes and/or color intensities
can improve the cognitive interaction as users will find it
easier to relate the animated tablets to their strengths, since
bigger and more intensely-colored tablets would be worth
more points. Furthermore, the games currently do not
account for the fact that users will gain competence over
time and probably stop playing. To improve users’
scalability of experience, splitting the games into varying
difficulty levels will continually challenge users and
provide a different experience each time they play the
games. Additional features to allow for customization of
the backgrounds and interfaces to suit users’ preferences,
or mounting the games on a variety of platforms like
PDAs, mobile phones, and social networking sites (e.g.
Facebook or MySpace) will not only facilitate expressive
interactions and co-experience, but also reinforce the
reflective and emotional threads of users’ overall
experiences.
3.3 The
Approach
Activity-Centered
Design
(ACD)
The ECD approach gives designers an insight to users’
experiences of the interaction tool. However, it does not
explain how the activity of playing these games affects the
user. Activity Theory (AT) describes a framework for
understanding how people operate in the world, taking
‘activity’ rather than ‘person’ or ‘mind’ as the central unit
of analysis [52-54]. Several other interpretations of AT exist,
but we will discuss the online tool based on the principles
described by Kaptelinin (Appendix 2) [53].
The principle on unity of consciousness and activity
states that the human mind (consciousness) is inseparable
from his interaction with the environment (activity) [52,53].
Users of the online tool know that the tablet colors in the
pill-catching game are related to their strengths, and the
objective of the hangman game is to learn about the
warfarin drug interactions. However, they may not see the
relevance of knowing the strengths and interactions. Thus,
providing a form of text or storyline would help make
9
users aware of the consequences of DRPs such as underand overdosing, and the severity of a drug interaction with
warfarin.
Object-orientedness, in this case, is to educate users on
the warfarin tablet strengths and drug interactions. In a
broad sense, the object in this principle need not be related
to physical objects, but includes socially/ culturally
defined properties as well [52,54]. Although the tool fulfils
its objectives, the significance of the activity itself can be
enhanced through making explicit to the user why it is
important to know about the tablet strengths and the
consequences of the drug interactions.
The hierarchical structure of activity is associated with a
tri-level scheme describing activities, actions and
operations which are oriented towards the goals and
motive of the whole activity [52-54]. This hierarchy differs
in patients and students playing the games. Students would
want to know the tablet strengths and drug interactions to
better prepare for exams, instead of improving their health.
Based on Leontiev’s principles [52], the relationship
between higher and lower objectives of a patient who
undergoes anticoagulant therapy and uses the online tool
is illustrated by Fig. 4. The smooth transition of conscious
actions to subconscious operations when playing the
games orients the user towards the objectives of learning
about warfarin. A breakdown, however, will disrupt the
user’s game playing activity, and may lead to
disorientation of the user or even frustration. An example
would be the shift in alphabet locations when the browser
is resized, resulting in the user trying to find out where to
click the alphabets.
The concept of internalization-externalization states that
our mental processes are derived from external actions
through the course of internalization, and is related to the
socio-cultural environment [52-54]. There is currently no
means of knowing whether the user has assimilated the
learning objectives of the games. Feedback mechanisms
such as short quizzes on simple warfarin interactions or
doses of different colored tablet combinations can be
incorporated so that the user is able to ‘internalize’ the
knowledge he has gained from the tool and ‘externalize’
this knowledge by correctly answering the questions.
The principle of tool mediation is the most significant
concept in AT, and it describes how a tool reflects the
accumulation and transmission of social knowledge, and
experiences of others who have tried to solve similar
problems before to make the tool more efficient [52-54].
Improvements of the ‘tools’ in the games would also
improve the users’ cognitive skills and knowledge on
warfarin. For example, a pill-box, cupped hand or a mouth
to simulate catching the warfarin tablets would better
mediate the process of how a patient takes the medication
in reality. Similarly, an animated form of the traditional
‘pen-and-paper’ hangman can probably provide a more
10
familiar and fun way of learning the warfarin drug
interactions.
Lastly, the principle of development is used to
understand how tools are developed into their existing
form [52-54]. The underlying concepts of why the games
were developed have been explained throughout the
various sections of this paper, but it can also be used to
further develop and improve the games. Voice reporting of
the user’s score status can improve his gameplay so that he
does not need to simultaneously focus on the rapidly
changing scores and correlating the strengths of the
different colored tablets. Similarly, having different
difficulty levels in the hangman game can also ease the
user’s learning curve.
Fig. 4 Hierarchy of objectives of a patient on anticoagulant therapy, and
how they are affected by socio-cultural factors.
3.4 Pharmaco-Cybernetics from an Ecological
Perspective
The Ecological Systems Theory by Urie Bronfenbrenner
describes how users interact with their immediate
environments (micro, meso, exo, macro, chrono), and how
these environments affect the user in a wider context [55].
From a pharmaco-cybernetics perspective, this theory can
be applied in the context of users learning about
anticoagulant therapy from the interaction tool (Appendix
3). The bi-directional influences of each individual system
on the others can help identify possible avenues for
improvement, as well as the pitfalls and disturbances in
the activity of using the tool. This warfarin tool also
allows the possibility of creating other larger-scale and
more complex interactive tools that will not only
encompass the magnitude of influences across the various
environments, but also reduce DRPs by empowering
patients with the appropriate drug knowledge so that they
can better participate in their therapies and management
strategies with their healthcare
ultimately improve their health.
professionals,
and
4. Conclusion
Developers of healthcare interactive tools often
overlook relevant user characteristics, tasks, preferences
and usability issues, thus resulting in systems or tools that
decrease productivity or simply remain unusable [56].
Medical tools need to be robust and easy to use in a wide
variety of environments [57]. Thus, healthcare applications
must be carefully crafted to ensure that they meet the
standards and models outlined by their target users.
The integration of interactive media and informatics
technologies with the WWW has enabled computational
tools to play an important role in pharma-culture. In this
paper, the concept of ‘pharmaco-cybernetics’ is introduced
through the creation of an interactive tool on oral
anticoagulation therapy. Interactivity was developed in the
form of two games for users to learn about warfarin tablet
strengths and drug interactions. Currently, this tool is
largely based on the principles of UCD and ECD.
However, the potential of incorporating the ACD
approach in the designing of this tool is definitely
attractive, and can lead to better quality healthcare tools
for other chronic medication therapies. Prototype sketches
of how the games can be improved in future versions are
provided in Fig. 5. It is hoped that these improved
versions will not only cater towards enhancing the user’s
experience, but also his interactions with the tool.
In conclusion, pharmaco-cybernetics can empower
patients with the appropriate knowledge regarding their
therapy so that they can better participate in the
management of their health. This can potentially help them
to adapt to any changes in their dietary habits and
lifestyles, as well as improve compliance, and ultimately,
improve the pharmaceutical care of patients who are on
anticoagulant therapy. Healthcare providers, patients and
developers of health information systems should realize
the importance and know the concepts and related
principles when designing for pharmaco-cybernetics
applications. However, understanding how users structure
their individual experiences, immediate environments, and
tasks is just the beginning when designing such products.
Designers should also take into account how external
forces such as socio-cultural and inter-personal factors
shape a user’s overall experience, attitude and goals in
using the applications, and through an ecological
perspective so as to cater the interactive tools for a wider
audience; as well as how they can be applied to the
designing of other pharmaco-cybernetics products
involving medication therapies.
[4]
[5]
[6]
[7]
[8]
[9]
[10]
[11]
Fig. 5 Prototype sketches of improved versions of the interactive tool
consisting of (a) the warfarin pill-catching game and (b) the warfarin
hangman game.
Acknowledgments
The authors would like to thank Asst. Prof. Timothy
Marsh, lecturer for the NM5206 module, and Ms. Cecilia
Chua from the Republic Polytechnic, Singapore, for their
support for the WarfarINT tool and contributing to the
success of this article.
[12]
[13]
[14]
[15]
References
[1]
[2]
[3]
L. H. Lee, M. C. Kong, Y. M. Wong, et al., Clinical
Pharmacy Practice Guidelines. Anticoagulation Warfarin, March 2006 ed., Singapore: Ministry of Health,
Singapore, 2006.
American Society of Hospital Pharmacists, “ASHP
statement on pharmaceutical care”, American Journal of
Hospital Pharmacy, Vol. 50, 1993, pp. 1720-1723.
E. H. Shortliffe, and A. M. Garber, “Training synergies
between medical informatics and health services research”,
Journal of the American Medical Informatics
Association, Vol. 9, No. 2, 2002, pp. 133-139.
[16]
[17]
[18]
[19]
11
D. R. Masys, P. F. Brennan, J. G. Ozbolt, et al., “Are
medical informatics and nursing informatics distinct
disciplines? The 1999 ACMI debate”, Journal of the
American Medical Informatics Association, Vol. 7, No.
3, 2000, pp. 304-312.
H. J. Lowe, E. C. Lomax, and S. E. Polonkey, “The World
Wide Web: a review of an emerging internet-based
technology for the distribution of biomedical information”,
Journal of the American Medical Informatics
Association, Vol. 3, No. 1, 1996, pp. 1-14.
S. Nettleton, “The emergence of e-scaped medicine?”,
Sociology, Vol. 38, No. 4, 2004, pp. 661-679.
G. Collste, “The internet doctor and medical ethics: ethical
implications of the introduction of the internet into medical
encounters”, Medicine, Health Care, and Philosophy,
Vol. 5, No. 2, 2002, pp. 121-125.
B. Chung, C. E. Corbett, B. Boulet, et al., “Talking
Wellness: a description of a community-academic
partnered project to engage an African-American
community around depression through the use of poetry,
film, and photography”, Ethnicity and Disease, Vol. 16,
No. 1 Suppl 1, 2006, pp. S67-S78.
U. Ritterfeld, and S. A. Jin, “Addressing media stigma for
people experiencing mental illness using an entertainmenteducation strategy”, Journal of Health Psychology, Vol.
11, No. 2, 2006, pp. 247-267.
D. Cantor, “Uncertain enthusiasm: the American Cancer
Society, public education, and the problems of the movie,
1921-1960”, Bulletin of the History of Medicine, Vol. 81,
No. 1, 2007, pp. 39-69.
S. E. Lederer, “Dark victory: cancer and popular
Hollywood film”, Bulletin of the History of Medicine,
Vol. 81, No. 1, 2007, pp. 94-115.
K. L. Taylor, J. L. Davis 3rd, R. O. Turner, et al.,
“Educating African American men about the prostate
cancer screening dilemma: a randomized intervention”,
Cancer Epidemiology, Biomarkers and Prevention, Vol.
15, No. 11, 2006, pp. 2179-2188.
G. Wallmyr, and C. Welin, “Young people, pornography,
and sexuality: sources and attitudes”, Journal of School
Nursing, Vol. 22, No. 5, 2006, pp. 290-295.
C. Jackson, J. D. Brown, and K. L. L'Engle, “R-rated
movies, bedroom televisions, and initiation of smoking by
white and black adolescents”, Archives of Pediatrics and
Adolescent Medicine, Vol. 161, No. 3, 2007, pp. 260-268.
J. D. Sargent, “Smoking in movies: impact on adolescent
smoking”, Adolescent Medicine Clinics, Vol. 16, No. 2,
2005, pp. 345-370.
Jean-Guilhem, “World Wide Web Consortium (W3C)”;
http://www.w3.org/ [accessed 3 July 2009].
Internet Society (ISOC), “The Internet Engineering Task
Force (IETF)”; http://www.ietf.org/ [accessed 3 July 2009].
J. Liaskos, and M. Diomidus, “Multimedia technologies in
education”, Studies in Health Technology and
Informatics, Vol. 65, 2002, pp. 359-372.
G. G. Moghaddam, and M. Moballeghi, “Human-computer
interaction: guidelines for web animation”, East/West
Interaction : An Online English Chinese Journal, 2006;
http://eprints.rclis.org/archive/00007823/ [accessed 3 July
2009].
12
[20] R. E. Weiss, D. S. Knowlton, and G. R. Morrison,
“Principles for using animation in computer-based
instruction: theoretical heuristics for effective design”,
Computers in Human Behavior, Vol. 18, No. 4, 2002,
pp. 465-477.
[21] A. Demirjian, and B. David, “Learning medical and dental
sciences through interactive multi-media”, Medinfo, Vol. 8
Pt 2, 1995, p. 1705.
[22] R. W. Pho, S. Y. Lim, and B. P. Pereira, “Computer
applications in orthopaedics”, Annals of the Academy of
Medicine Singapore, Vol. 19, No. 5, 1990, pp. 691-698.
[23] K. Penska, L. Folio, and R. Bunger, “Medical applications
of digital image morphing”, Journal of Digital Imaging,
Vol. 20, No. 3, 2007, pp. 279-283.
[24] D. M. Smith, S. J. Aston, C. B. Cutting, et al.,
“Applications of virtual reality in aesthetic surgery”,
Plastic and Reconstructive Surgery, Vol. 116, No. 3,
2005, pp. 898-904.
[25] D. M. Smith, S. J. Aston, C. B. Cutting, et al., “Designing
a virtual reality model for aesthetic surgery”, Plastic and
Reconstructive Surgery, Vol. 116, No. 3, 2005, pp. 893897.
[26] R. C. Ward, M. W. Yambert, R. J. Toedte, et al., “Creating
a human phantom for the virtual human program”, Studies
in Health Technology and Informatics, Vol. 70, 2000,
pp. 368-374.
[27] J. E. Champoux, “Animated films as a teaching resource”,
Journal of Management Education, Vol. 25, No. 1, 2001,
pp. 79-100.
[28] P. Piyasirivej, “Towards usability evaluation of Flash web
sites”, in World Forum Proceedings of the International
Research Foundation for Development, Tunis, Tunisia,
2005.
[29] Edheads + COSI, “Virtual Knee Surgery and Choose the
Prosthetic games”; http://www.edheads.org/activities/knee/
[accessed 3 July 2009].
[30] H. Leufkens, Y. Hekster, and S. Hudson, “Scenario
analysis of the future of clinical pharmacy”, Pharmacy
World and Science, Vol. 19, No. 4, 1997, pp. 182-185.
[31] T. Westerlund, A. B. Almarsdóttir, and A. Melander,
“Factors influencing the detection rate of drug-related
problems in community pharmacy”, Pharmacy World
and Science, Vol. 21, No. 6, 1999, pp. 245-250.
[32] T. L. Stedman, Stedman's Medical Dictionary, 28th ed.,
Baltimore, Maryland: Lippincott Williams & Wilkins,
2005.
[33] P. Pangaro, “"Getting started" guide to cybernetics”;
http://pangaro.com/published/cyber-macmillan.html
[34] Wikipedia, “Cybernetics”;
http://en.wikipedia.org/wiki/Cybernetics [accessed 3 July
2009].
[35] American Society for Cybernetics, “Foundations - the
subject of cybernetics: defining 'cybernetics'”;
http://www.asc-cybernetics.org/foundations/definitions.htm
[36] P. A. Corning, “Synergy, cybernetics, and the evolution of
politics”, International Political Science Review, Vol. 17,
No. 1, 1996, pp. 91-119.
[37] 'Psycho-cybernetics' Author, “Plastic surgeon tries to heal
'inner scars'”, Los Angeles TimesB5, 1973, pp. B5-1.
[38] Wikipedia, “New cybernetics”;
http://en.wikipedia.org/wiki/New_cybernetics [accessed 3
July 2009].
[39] F. Geyer, “The challenge of sociocybernetics”,
Kybernetes: The International Journal of Systems &
Cybernetics, Vol. 24, No. 4, 1995, pp. 6-32.
[40] P. C. Adams, “Cyberspace and virtual places”,
Geographical Review, Vol. 87, No. 2, 1997, pp. 155-171.
[41] R. Trappl, “The cybernetics and systems revival”, in 14th
European Meeting on Cybernetics and Systems
Research (EMCSR'98), University of Vienna, Austrian
Society for Cybernetic Studies, February 1998, p. Preface.
[42] W. Gibson, Neuromancer (Special 20th Anniversary
Edition), Hardcover ed.: Ace Books, 2004.
[43] Wikipedia, “The Matrix”;
http://en.wikipedia.org/wiki/Matrix_movie [accessed 3 July
2009].
[44] Wikipedia, “Cyberspace”;
http://en.wikipedia.org/wiki/Cyberspace [accessed 3 July
2009].
[45] G. R. S. Weir, and S. Heeps, “Getting the message across:
ten principles for web animation. ”;
http://eprints.cdlr.strath.ac.uk/2513/ [accessed 3 July
2009].
[46] D. A. Norman, “The psychopathology of everyday things”,
The Design of Everyday Things, pp. 1-33, USA: Perseus
Publishing, 2002.
[47] J. Preece, Y. Rogers, and H. Sharp, Interaction design:
beyond human-computer interaction, 1st ed., New York,
NY: John Wiley & Sons Inc., 2002.
[48] D. A. Norman, “Emotion and design: attractive things work
better”, Interactions, Vol. 9, No. 4, 2002, pp. 36-42.
[49] D. A. Norman, “Three levels of design: visceral,
behavioral, and reflective”, Emotional Design: Why We
Love (or Hate) Everyday Things, pp. 63-98, New York,
NY: Basic Books, 2004.
[50] J. McCarthy, and P. Wright, “The threads of experience”,
Technology as Experience, pp. 79-104, Cambridge, MA:
MIT Press, 2004.
[51] J. Forlizzi, and K. Battarbee, “Understanding experience in
interactive systems”, Symposium on Designing
Interactive Systems. Proceedings of the 5th Conference
on Designing Interactive Systems: Processes, Practices,
Methods, and Techniques. DIS '04, 2004, pp. 261-268;
http://doi.acm.org/10.1145/1013115.1013152 [accessed 3
July 2009].
[52] V. Kaptelinin, K. Kuutti, and L. Bannon, “Activity theory:
basic concepts and applications. A summary of a tutorial
given at the east west HCI95 conference”, Lecture Notes
in Computer Science, Human-Computer Interaction,
Vol. 1015, 1995, pp. 189-201.
[53] V. Kaptelinin, “Activity theory: implications for humancomputer interaction”, Context and Consciousness:
Activity Theory and Human-Computer Interaction, B.
A. Nardi, ed., pp. 103-116, Cambridge, MA: MIT Press,
1996.
[54] V. Kaptelinin, and B. A. Nardi, “Activity theory in a
nutshell”, Acting with Technology: Activity Theory and
Interaction Design, pp. 29-72, Cambridge, MA: MIT
Press, 2006.
[55] D. Paquette, and J. Ryan, “Bronfenbrenner’s Ecological
Systems Theory”;
http://pt3.nl.edu/paquetteryanwebquest.pdf [accessed 3
July 2009].
[56] C. M. Johnson, T. R. Johnson, and J. Zhang, “A usercentered framework for redesigning health care interfaces”,
Journal of Biomedical Informatics, Vol. 38, No. 1, 2005,
pp. 75-87.
[57] R. Gagnier, “User centered design of medical devices:
managing use related hazards”, Macadamian White
Papers, n.d.; http://www.macadamianusability.com/resources/whitepapers/Maskerydesigning_medical_devices.pdf [accessed 10 November
2008].
Kevin Y.-L. Yap (B.Sc. in Pharmacy (Hons), M.Eng., Sp. Dip.
Digital Media Creation) is currently a Ph.D. candidate in the
Department of Pharmacy, National University of Singapore, and a
registered pharmacist in Singapore. He has worked as a
pharmacist in the hospital and community settings, as well as an
academic facilitator in the biomedical sciences, based on the
problem-based learning pedagogy. His research interests lie in the
application of informatics, digital media, interactive and web
technologies in clinical pharmacy practice, particularly with regards
to pharmaceutical care and the solving of drug-related problems;
and he has presented in various international conferences and
published several papers in this area. He is a member of the
Pharmaceutical Society of Singapore, American Association for
the Advancement of Science, and the Healthcare Information and
Management Systems Society. He has also been featured in
th
Marquis Who’s Who in Science and Engineering (10 ed.), and in
th
Medicine and Healthcare (7 ed.).
Xuejin Chuang, Alvin J.M. Lee, Raemarie Z. Lee, Lijuan Lim
and Jeanette J. Lim were undergraduates, while R. Nimesha and
Kevin Yap were postgraduates in the National University of
Singapore during the time in which the pilot usability study was
carried out. The WarfarINT tool was originally designed and
created by Kevin Yap. All authors were members of the project
team in the module NM5206 Emerging Media Interaction Design
offered by the Communications and New Media (CNM)
Programme, Faculty of Arts and Social Science, in the first
semester of the Academic Year 2008-2009.
13
14
ISSN (Print): 1694-0814
SIMILARITY MATCHING TECHNIQUES FOR FAULT
DIAGNOSIS IN AUTOMOTIVE INFOTAINMENT
ELECTRONICS
Dr. Mashud Kabir
Department of Computer Science, University of Tuebingen
D-72027 Tuebingen, Germany
Abstract
Fault diagnosis has become a very important area of research
during the last decade due to the advancement of mechanical and
electrical systems in industries. The automobile is a crucial field
where fault diagnosis is given a special attention. Due to the
increasing complexity and newly added features in vehicles, a
comprehensive study has to be performed in order to achieve an
appropriate diagnosis model. A diagnosis system is capable of
identifying the faults of a system by investigating the observable
effects (or symptoms). The system categorizes the fault into a
diagnosis class and identifies a probable cause based on the
supplied fault symptoms. Fault categorization and identification
are done using similarity matching techniques. The development
of diagnosis classes is done by making use of previous
experience, knowledge or information within an application area.
The necessary information used may come from several sources
of knowledge, such as from system analysis. In this paper
similarity matching techniques for fault diagnosis in automotive
infotainment applications are discussed.
Key words: similarity, fault, diagnosis, matching, automotive,
infotainment, cosine.
1. Introduction
At first feature selection is discussed where stop word list
and word stemming are used. Then pattern recognition is
explained. Ranking algorithms are used to rank words,
web pages. Page ranking algorithm is discussed keeping in
mind our application. Similarity algorithms are discussed
in the next sections. Then proper similarity matching
algorithm which best fits to fault diagnosis in automotive
infotainment system is presented. The algorithm is
analyzed with real field data and the results are evaluated.
2. Feature Selection
Feature selection is one of the main steps in similarity
matching of faults. We apply a stop word [1] list to filter
out the meaningless words. A list of stop words has been
built. This list has been created keeping in mind the
existing standard fault description language in automotive
infotainment systems.
Word stemming is a method where lexically similar words
are listed together. Here, the words with affixes and
suffixes are converted into root words. This methodology
overcomes the limitation of words with the same meaning
being categorized into different classes. A list of stemming
words is created for automotive infotainment system. Both
the lists of stop words and stemming words were
developed with the help of experienced system engineers
in automotive infotainment system.
3. Pattern Recognition
Based on highly developed skill after sensing the
surroundings, humans are capable of taking any actions
according to their observations. By observing the nature of
human intelligence, a machine can be built to do the same
job, such as identifying hand writing, post code, voice,
finger print, DNA, human face etc.
A pattern is an abstract object such as a set of
measurements describing a physical object. This is an
entity with a given name such as hand writing, a sentence,
human face etc. Pattern recognition consists of several
steps such as observation of inputs, learning how to
distinguish different patterns and making rationale
decisions in categorizing patterns.
Shmuel Brody [3] has summarized the concepts of pattern
recognition and their uses in similarity matching. Human
detected patterns contain many relevant and irrelevant
data. The most important task in pattern recognition is to
find out the meaningful patterns and to disregard the
irrelevant subject matter. The fields of area of pattern
recognition range from data analysis, feature extraction,
error estimation, error removal,
grammatical inference and parsing.
cluster
analysis,
Faramarz Valafar [4] has discussed pattern recognition
techniques in data analysis. Clustering is one of the most
commonly used recognition techniques. Data are grouped
into clusters or groups in clustering. K-means clustering
[2] is a widely used algorithm for data clustering. In kmeans similar algorithm patterns are partitioned into the
same group. All the data are classified into any of the k
clusters or classes. Then the mean inter and intra-class
distances are determined. The last step is to maximize the
intra-class distance and minimize the inter-class distance.
This is an iterative procedure where data is moved from
one cluster to another. This process continues until
optimized distances of intra-class and inter-class are
found.
In pattern
similarity
discover
similarity
detecting.
15
recognition different techniques are applied for
matching. For this work it is necessary to
optimized techniques and algorithms for
matching, fault classifying and fault cause
Fig. 1 Inbound link of page A.
A set of five web pages is assumed: A, B, C, D, E. The
initial probability is distributed evenly among these pages.
Therefore, each of the pages will get a PageRank of 1.0/5.
It means,
PR(A) = PR(B) = PR(C) = PR(D) = PR(E) = 0.2
(i)
Now suppose the scenario as depicted in figure 1:
Page A has inbound links from Page C, D and E. Thus, the
PageRank of page A
PR(A) =PR(C) + PR(D) + PR(E)
Page C has other outbound links to page E, page D has
other outbound links to B, C and E as depicted in figure 2.
E
C
4. Ranking Algorithms
PageRank algorithm is a widely used algorithm to rank
web pages according to their importance. The algorithm is
described as following –
PageRank is a link analysis algorithm to rank a web page
from a set of pages according to its relative importance. It
provides a numerical weighting to each of the page
elements in the set. This weighting is called PageRank of
E which is denoted by PR(E).
PageRank was introduced by Larry Page at Stanford
University to develop a new search engine in the web. The
ranking of a page depends on the number of links of the
other pages to that page.
PageRank is a probability distribution which shows the
likelihood that a user randomly clicking on the links finds
a specific site. This probability ranges from 0 to 1. A
PageRank of 0.8 means that the probability of reaching a
specific site by randomly clicking on a set of links is 80%.
(ii)
E
D
C
B
Fig. 2 The Outbound links of page C and Page D.
The value of the link-votes is divided among all the
outbound links of a page. Thus, page C contributes a vote
weight of 0.2/2 i.e. 0.1 and page D contributes a vote
weight of 0.2/4 i.e. 0.05.
Thus, the equation stands in the following form:
PR( A) =
PR(C ) PR( D) PR( E )
+
+
2
4
1
(iii)
The above equation can be generalized in the following
form assuming that the PageRank incurred by an outband
link of a page is the page’s own PageRank in the set
divided by the number of outband links
PR( A) =
PR(C ) PR( D) PR( E )
+
+
L(C )
L( D)
L( E )
(iv)
D
The PageRank of any page i can be expressed in the
following form:
C
A
E
16
ISSN (Print): 1694-0814
PR(i ) =∑
jεS i
PR( j )
Nj
(v)
Where,
PR(i ) = PageRank of page i
PR( j ) = PageRank of any other pages except
page i.
N j = Number of pages in the set
jεS i = Inbound pages linking to page i
PageRank algorithm is mainly used for internet
applications to find the rank of a page. The basis of the
algorithm is that the rank of a page depends on the
inbound links of the other pages. To apply this technique
we need to compare links among the pages with the links
among the features of the fault. But this study requires the
ranking of features according to their importance. This
makes PageRank algorithm inappropriate for this project.
5.
Similarity Matching
This chapter describes the similarity matching techniques
for strings. Using these techniques, a concept is proposed
to search similar faults when the symptoms of a fault are
provided.
Edit distance is a common term in matching algorithms.
The word distance is used to compare different data for
similarity. Edit distance is a measure to estimate
differences between input elements. Different methods to
calculate edit distance exist:
Levenshtein Distance
Levenshtein distance is named after the Russian scientist
Vladimir Levenshtein, who devised the algorithm in 1965.
The Levenshtein distance between two strings is given by
the minimum number of operations needed to transform
one string into the other, where an operation is an
insertion, deletion, or substitution of a single character.
Levenshtein distance (LD) is a measure of the similarity
between two inputs: the source s and the target input t.
The distance is the number of deletions, insertions, or
substitutions required to transform s into t. For example,
If s is "math" and t is "math", then LD(s,t) = 0, because no
transformations are needed.
If s is "math" and t is "mats", then LD(s,t) = 1, because
one substitution (change "h" to "s") is sufficient to
transform s into t.
The more different the inputs are, the greater the
Levenshtein distance is.
Insertion, deletion and substitution are the main criteria for
determining Levenshtein Distance. The position of a
character plays an important role to determine the
distance. In this study, the description of a fault is dealt
with. If Levenshtein Distance is applied to find out the
similarity of faults it would not give a meaningful result as
the positions of the strings should not have importance.
That is why this technique will not be used in this study.
Damerau-Levenshtein Distance
Damerau-Levenshtein distance comes from Levenshtein
distance that counts transposition as a single edit
operation. The Damerau-Levenshtein distance is equal to
the minimal number of insertions, deletions, substitutions
and transpositions needed to transform one string into the
other.
Kukich [5] described several edit distance algorithms
which use Damerau-Levenshtein distance. It has been
proved that the use of Damerau-Levenshtein metric to
calculate the similarity between two words is a slow
process. For this reason this method is not well-suited for
similarity matching in this project.
Needleman – Wunsch Distance
The Levenshtein distance algorithm assumes that the cost
of all insertions, deletions, substitutions or conversions is
equal. However, in some scenarios this may not be
desirable and may mask the acceptable distances between
inputs.
Needleman-Wunsch has modified Levenshtein distance
algorithm to add cost matrix as an extra input. This matrix
structure contains two cost matrics for each pair of
characters to convert from and to. The cost of inserting
this character and converting between characters is listed
in this matrix.
This approach is not appropriate for use in this study’s
similarity matching for the same reason stated in
Levenshtein approach.
Hamming Distance
The Hamming distance [6] H is defined for the same
length inputs. For two inputs s and t, H(s, t) is the
number of places in which the two strings differ, i.e., have
different characters.
Hamming Distance is used in information theory. This
method can not be applied in similarity matching for
automotive faults since Hamming Distance only considers
the differences among the two inputs.
17
Weighted Edit Distance
This algorithm differs from the Edit Distance in
weighting. A particular weight is imposed for each
operation of insertion, deletion and substitution.
The main goal of similarity matching of faults is to find
the faults with the similar behaviors. Weighted Edit
Distance focuses on providing weight on the operations.
This kind of approach is inappropriate for finding similar
faults.
Hamming Distance
The Hamming distance is the number of positions for
which the corresponding characters differ. It is simply the
number of differences between two strings of the same
length. For example:
The Hamming Distance between GERMANY and
IRELAND is 5.
To apply this distance between two error features they
must be of equal length, which is a rare case. This results
in the decision not to use Hamming Distance for similarity
matching in this study.
Cos similarity ( A, B ) =
∑
∑
N
i =1
N
i =1
α i Fi ( A) Fi ( B )
α i 2 Fi ( A) 2 ∑i =1α i 2 Fi ( B ) 2
N
where,
α i = user-determined parameter (weights) (~1)
Cosine similarity method counts the number of different
words in two documents. With this method the highest
frequency words within any document will have the
largest influence on its similarity with other documents.
Documents with many occurrences of an unusual word or
many different unusual words will have low cosine
similarity measures with most other documents. Weighting
schemes are frequently used to modify the standard cosine
measure. These typically lower the importance of common
words.
Below are the results of some input data and their
similarities with the existing input database using this
algorithm –
Input Database:
6. Similarity Determination
The aim of this section is to propose an algorithm to use
for similarity matching in text queries. The procedures of
this algorithm are as following
A text (query) T is represented by multidimensional
vector:
F(T) = (F1(T), F2(T), …Fk(T)) (occurrence vector)
k = no. of distinct term occurring in database (non-stop
word)
Function of frequency of the i-th term in T,
tf i
1 ⎛⎜
1+
⎜
2 ⎝ max tf i T
T
Fi (T ) =
⎞
⎟ log N
⎟
ni
⎠
where,
tf i
T
= frequency of the i-th term in T
max tf i
T
= no. of database documents where the most
frequent term of T occurs
N = no. of database entries
ni
= no. of entries where the i-th term occurs
The cosine similarity measure between a query (A) and a
stored document (B) is defined as:
This is the database which is already stored in the system.
This is compared with the user provided fault symptoms.
Attachment
Y
message
Defect ID
Fault Characteristics
32
Display ON Signal will be
sent, but Display remains dark
40
Preconditions: radio hu
41
audio hu radio message message
42
hu -> audio
Y
44
message message radio radio
45
message
46
message
47
message no sds
48
no message sds
49
50
51
52
53
message from headunit
radio:
radio.
Preconditions:
message;
message->
preconditions: hu sds sdars
radio hu message message
radio
hu
no
message
radio
no
hu
message
radio hu message message
radio hu
radio
radio radio
radio radio radio
radio does not
receive
Result Analysis:
Below is the graphical representation of outputs for
determining fault similarities corresponding to user
18
ISSN (Print): 1694-0814
radio hu
radio dvd message
80
similarity
provided fault symptoms. The similarities of fault
symptom radio hu are 100% (fault id 49) with the database
fault radio hu and 68% (fault id 40) with the database fault
Preconditions: radio hu message. The result of this fault
matching is shown in figure 3.
60
40
20
0
40 41 42 44 45 46 47 48 49 50 51 52 53
similarity %
100
fault id
80
60
40
Fig. 5 Fault similarities with symptom radio dvd message.
20
0
40 41 42 44 45 46 47 48 49 50 51 52 53
fault id
Fig. 3 Fault similarities with symptom radio hu.
The similarities of fault symptom radio hu message are
100% (id 40) with database fault radio hu and 84% (id 41)
with database fault radio: radio. message; audio hu radio
message message and 84% (id 45) with database fault
radio hu message message messaeg. The result of this fault
matching is shown in figure 4.
The similarities of fault symptom radio dvd message are
61% (id 41) with the database fault radio: radio. message;
audio hu radio message message and 56% (id 45) with the
database fault radio hu message message message. The
result of this fault matching is shown in figure 5.
The similarities of fault symptom radio dvd are 58% with
database faults radio (id 50) and radio radio(id 51) and
radio radio radio (id 52). The result of this fault matching
is shown in figure 6.
radio dvd
radio hu message
similarity %
100
80
similarity %
100
80
60
40
20
0
60
40
40 41 44 45 46 47 48 49 50 51 52 53
20
fault id
0
40 41 42 44 45 46 47 48 49 50 51 52 53
Fig. 6 Fault similarities with symptom radio dvd.
fault id
Fig. 4 Fault similarities with symptom radio hu message.
Based on the above result analysis it can be concluded that
the similarity of a user provided fault is higher if the
symptom of the fault matches more closely with any
database fault. It satisfies the requirement of finding
similar faults for a fault symptom.
7. Conclusion
In this paper feature selection, pattern recognition, page
ranking algorithms have been discussed to process input
data and system database. Different similarity matching
algorithms have also been explained. Cosine similarity
algorithm has been chosen for our special application of
automotive infotainment system. Real field automotive
faults have been used to analyse the cosine similarity
method. After comparing with the existing fault database,
a decision of fault similarities on user provided fault has
been made and the results have been discussed.
References
[1] Eric Brill, Jimmy Lin, Michele Banko, Susan Dumais,
Andrew Ng. Data-Intensive Question-Answering. In the
proceedings of the Tenth Text Retrieval Conference (TREC
2001), Maryland,November 2001.
[2] Tapas Kanungoy, David Mountz, Nathan Netanyahu,
Christine Piatko, Ruth Silverman, Angela Wu. A Local Search
Approximation Algorithm for k-Means Clustering. 18th Annual
ACM Symposium on Computational Geometry (SoCG’02),
Barcelona, Spain, June 2002.
[3] Shmuel Brody. Cluster-Based Pattern Recognition in Natural
Language Test. Master Thesis. August 2005.
[4] Faramarz Valafar. Pattern Recognition Techniques in
Microarray Data Analysis: A Survey. Special issue of Annals of
New York Academy of Sciences, Techniques in
Bioinformatics and Medical Informatics. (980) 41-64,
December 2002.
[5] Karen Kukich. Techniques for automatically correcting
words in text. ACM Computing Surveys, 24(4):377–439,
December 1992.
[6] Yu Tao; Muthukkumarasamy, V.; Verma, B.; Blumenstein,
M. A texture extraction technique using 2D-DFT and Hamming
distance. Fifth International Conference on Computational
Intelligence and Multimedia Applications, 2003. ICCIMA
2003.
Mashud Kabir. I was born in Narayanganj, Bangladesh in 1976. I have
completed my Bachelor of Science (BSc) in Electrical & Electronic
Engineering from Bangladesh University of Engineering & Technology
(BUET) in 2000. I was awarded board scholarship from 1995 to 2000. I
earned my Master of Science (MSc) in Communication Engineering from
University of Stuttgart, Germany in 2003. I achieved STIEBET German
Government scholarship during my Master Study. My Master thesis was
“Region-Based Adaptation of Diffusion Protocols in MANETs” where up
to 21% of broadcast can be saved. I worked at Mercedes-Benz
Technology Center, Germany from 2003 to 2005 as a PhD student. I have
worked in the research & development projects of BMW, Land-Rover
and Audi in Automotive Infotainment Network area for more than four
years. I have achieved my Doctoral degree from the Department of
Computer Science, University of Tuebingen, Germany in 2008. My
dissertation topic was “Intelligent System for Fault Diagnosis in
Automotive Applications”.
19
20
ISSN (Print): 1694-0814
Prototype System for Retrieval of Remote Sensing Images based
on Color Moment and Gray Level Co-Occurrence Matrix
1
Priti Maheshwary1 and Namita Sricastava2
Deparment of Computer Application, Maulana Azad National Institute of Technology
Bhopal, Madhya Pradesh, India
2
Deparment of Mathematics, Maulana Azad National Institute of Technology
Bhopal, Madhya Pradesh, India
Abstract
The remote sensing image archive is increasing day by day. The
storage, organization and retrieval of these images poses a
challenge to the scienitific community. In this paper we have
developed a system for retrieval of remote sensing images on
the basis of color moment and gray level co-occurrence matrix
feature extractor. The results obtained through prototype system
is encouraging.
Key words: Remote Sensing Image Retrieval, Color Moment,
Gray Level Co-occurrence Matrix, Clustering index.
1. Introduction
Content-based image retrieval (CBIR) technology was
proposed in 1990s and it is an image retrieval technology
using image vision contents such as color, texture, shape,
spatial relationship, not using image notation to search
images. It resolves some traditional image retrieval
problems, for example, manual notations for images bring
users a large amount of workload and inaccurate
subjective description. After more than one decade, it has
been developed as content-based vision information
retrieval technology including image information and
video information. Great progress has been made in theory
and applications.
At present, CBIR technology obtains successful
applications in face reorganization fields, fingerprint
reorganization fields, medical image database fields,
trademark registration fields, etc., such as QBIC system of
IBM Corporation, Photobook system of MIT Media
Laboratory and Virage system of Virage Corporation. It is
difficult to apply these systems in massive remote sensing
image archive because remote sensing image has many
features including various data types, a mass of data,
different resolution scales and different data sources,
which restrict the application of CBIR technology in
remote sensing image field. In order to change the current
situation, we must resolve some problems as follows.
1) Storing massive remote sensing image data.
2) Designing reasonable physical and logical pattern of
remote sensing image database.
3) Adopting adaptive image feature extraction algorithms.
4) Adopting indexing structure for search.
5) Designing reasonable content based searching system
of massive remote sensing image database.
The rest of the paper is arranged as follows. In Sec. 2, we
discuss the methodology. In Sec. 3, the experimental setup
and the results obtained are discussed. We conclude in
Sec. 4.
2. Methodology
For practical applications, users are often interested in
the partial region or targets, such as military target,
public targets and ground resource targets in remote
sensing image instead of the entire image. For example,
the small scale important targets and regions of remote
sensing image arrest more attention than the entire
remote sensing image in application. These image slice
features of important targets and regions extracted by
color, texture, shape, spatial relationship, etc. are stored
in feature database. Efficient indexing technology is a
key factor for applying the content-based image
retrieval in massive image database successfully.
Indexing technology developed from traditional
database and has been applied in content-based image
retrieval field subsequently. Fig. 1 shows an
architecture frame of content-based remote sensing
image.
Traditionally, satellite image classification has been done
at the pixel level. For a typical LISS III image has 23.5m
resolution, a 100 × 100 sized image patch covers roughly
7.2 Km2. This is too large an area to represent precise
ground segmentation, but our focus is more on building a
querying and browsing system than showing exact
boundaries between classes. Dividing the image into
rectangular patches makes it very convenient for training
as well as browsing. Since users of such systems are
generally more interested in getting an overview of the
location, zooming and panning is allowed optionally as
part of the interface.
21
Skewness can be understood as a measure of the degree of
asymmetry in the distribution.
2.2 Grey-level co-occurrence matrix texture
Grey-Level Co-occurrence Matrix texture measurements
have been the workhorse of image texture since they were
proposed by Haralick in the 1970s. To many image
analysts, they are a button you push in the software that
yields a band whose use improves classification - or
not. The original works are necessarily condensed and
mathematical, making the process difficult to understand
for the student or front-line image analyst.
Calculate the selected Feature. This calculation uses only
the values in the GLCM. See:
i) Contrast
Figure 1: Architectural Framework of CBIR system
We have developed a prototype system for image
retrieval. In this a query image is taken and images similar
to the query images are found on the basis of color and
texture similarity. The three main tasks of the system are:
ii) Correlation
iii) Energy
iv) Homogeneity
1.
2.
3.
4.
Color Moment Feature Extraction
GLCM Texture Feature Extraction.
K-means clustering to form index.
Retrieval between the query image and database.
These features are calculated with distance 1 and angle 0,
45 and 90 degrees.
2.1 Color moment:
We will define the ith color channel at the jth
image pixel as pij. The three color moments can
then be defined as:
MOMENT 1 – Mean:
Mean can be understood as the average color value in the
image.
MOMENT 2 -Standard Deviation:
The standard deviation is the square root of the variance of
the distribution.
MOMENT 3 – Skewness:
2.3 K-Means Clustering
A cluster is a collection of data objects that are similar to
one another with in the same cluster and are dissimilar to
the objects in the other clusters. It is the best suited for
data mining because of its efficiency in processing large
data sets. It is defined as follows:
The k-means algorithm is built upon four basic operations:
1. Selection of the initial k-means for k-clusters.
2. Calculation of the dissimilarity between an object
and the mean of a cluster.
3. Allocation of an object of the cluster whose mean
is nearest to the object.
4. Re-calculation of the mean of a cluster from the
object allocated to it so that the intra cluster
dissimilarity is minimized.
The advantage of K-means algorithm is that it works well
when clusters are not well separated from each other,
which is frequently encountered in images. The cluster
ISSN (Print): 1694-0814
number allotted to each image is considered its class or
group.
2.4 Similarity Matching:
The four major classifications of images are shown in
figure 2 to 5. Figure 6 and 7 shows the content based
retrieval system. We get 80% to 83% accuracy in our
results.
Many similarity measures have been developed for image
retrieval based on empirical estimates of the feature
extraction. We have used Euclidean Distance for
similarity matching.
The Euclidean distance between two points P = (p1, p2,
pn) and Q = (q1,q2, ……, qn), in Euclidean n-space
defined as:
……,
Now for the retrieval purpose the user select the query
patch and on the basis of its class number the distance
between the query patch with the other images of that
class is calculated and images are retrieved.
Figure 2: Water bodies
3. Experimental Plan
For our experiments, we use 3 LISS III + multi-spectral
satellite images with 23.5m resolution. We choose to
support 4 semantic categories in our experimental system,
namely mountain, water bodies, vegetation, and residential
area. In consultation with an expert in satellite image
analysis, we choose near-IR (infra-red), red and green
bands as the three spectral channels for classification as
well as display. The reasons for this choice are as follows.
Near-IR band is selected over blue band because of a
somewhat inverse relationship between a healthy plant’s
reflectivity in near-IR and red, i.e., healthy vegetation
reflects high in near-IR and low in red. Near-IR and red
bands are key to differentiating between vegetation types
and states. Blue light is very abundant in the atmosphere
and is diffracted all over the place. It therefore is very
noisy. Hence use of blue band is often avoided. Visible
green is used because it is less noisy and provides unique
information compared to Near IR and red. The pixel
dimensions of each satellite image are used in our
experiments are 720x540, with geographic dimensions
being approximately 51.84Km× 38.88Km. The choice
patch size is critical. A patch should be large enough to
encapsulate the visual features of a semantic category,
while being small enough to include only one semantic
category in most cases. We choose patch size 100×100
pixels. We obtain 80 patches from all the images in this
manner. These patches are stored in a database along with
the identity of their parent images and the relative location
within them. Ground truth categorization is not available
readily for our patches.
Figure 3: Open Land with vegetation
Figure 4: Buildings
22
23
content and knowledge base for finding vegetation or
water or building areas.
5. References
Figure 5: Vegetation and Mountain
Figure 6: CBIR System
Figure 7: Screen 2 of CBIR System
4. Conclusions
For retrieving similar images to a given query image we
have developed a prototype system. We get fruitful results
on the example images used in the experiments. We can
use this technique for mining similar images based on
[1] Li, J., Wang, J. Z. and Wiederhold, G., “Integrated
Region Matching for Image Retrieval,” ACM Multimedia,
2000, pp. 147-156.
[2] Flickner, M., Sawhney, H., Niblack, W., Ashley, J.,
Huang, Q., Dom, B., Gorkani, M., Hafner, J., Lee, D.,
Petkovic, D., Steele, D. and Yanker, P., “Query by image
and video content: The QBIC system,” IEEE Computer,
28(9), 1995, pp. 23-32
[3] Pentland, A., Picard, R. and Sclaroff S., “Photobook:
Contentbased manipulation of image databases”,
International Journal of Computer Vision, 18(3), 1996,
pp. 233–254
[4] Smith, J.R., and Chang, S.F., “Single color extraction
and image query,” In Proceeding IEEE International
Conference on Image Processing, 1997, pp. 528–531
[5]Gupta, A., and Jain, R., “Visual information retrieval,”
Comm. Assoc. Comp. Mach., 40(5), 1997, pp. 70–79
[6]Eka Aulia, “Heirarchical Indexing for Region based
image retrieval”, A thesis Submitted to the Graduate
Faculty of the Louisiana State University and Agricultural
and Mechanical College.
[7]Shi, J., and Malik, J., “Normalized Cuts and Image
Segmentation,” Proceedings Computer Vision and Pattern
Recognition, June, 1997, pp. 731-737
[8]Smith, J., “Color for Image Retrieval”, Image
Databases: Search and Retrieval of Digital Imagery, John
Wiley & Sons, New York, 2001, pp. 285-311
[9]Zhang, R. and Zhang, Z., (2002), “A Clustering Based
Approach to Efficient Image Retrieval,” Proceedings of
the 14th IEEE International Conference on Tools with
Artificial Intelligence, pp. 339
24
ISSN (Print): 1694-0814
Performing Hybrid Recommendation in Intermodal Transportation –
the FTMarket System’s Recommendation Module
Alexis Lazanas
Industrial Management and Information Systems Lab, University of Patras
Rion Patras, 26500, Greece
Abstract
Diverse recommendation techniques have been already proposed
and encapsulated into several e-business applications, aiming to
perform a more accurate evaluation of the existing information
and accordingly augment the assistance provided to the users
involved. This paper reports on the development and integration
of a recommendation module in an agent-based transportation
transactions management system. The module is built according
to a novel hybrid recommendation technique, which combines
the advantages of collaborative filtering and knowledge-based
approaches. The proposed technique and supporting module
assist customers in considering in detail alternative transportation
transactions that satisfy their requests, as well as in evaluating
completed transactions. The related services are invoked through
a software agent that constructs the appropriate knowledge rules
and performs a synthesis of the recommendation policy.
Key words: Data mining, Knowledge Association Rules,
Recommender systems, Intermodal Transportation.
1. Introduction
Transportation management involves diverse decision
making issues, which are basically related to the
appropriate route and carrier selection. Such issues mainly
raise due to the variety of the customer’s preferences (e.g.
cost limitations, loading preferences, delivery dates) and
the carrier’s service resources (e.g. transportation media,
available itineraries, capacity). The matching between the
above preferences and offered services cannot be easily
handled manually, as in most cases a plethora of
alternative options exist, while time and money limitations
are ubiquitous. Generally speaking, transportation
transactions management requires quick and cost-effective
solutions to the customers’ demands for both distribution
and shipping operations. In cases where many alternatives
exist, there is an urgent need for providing
recommendations. The customer should be assisted in
order to properly evaluate the proposed alternatives and
make his/her final decision.
Recommendation systems have been described as
systems that produce individualized recommendations or
have the effect of guiding the user in a personalized way,
in environments where the amount of on-line information
vastly outstrips any individual’s capability to survey it [2].
Generally speaking, such systems represent the users’
preferences for the purpose of submitting suggestions for
purchasing or evaluating elements. Fundamental
applications can be found in the fields of electronic
commerce and information retrieval, where they provide
suggestions that effectively direct the users to the elements
that satisfy better their necessities and preferences [21].
This paper reports on the development of an
innovative recommendation module that provides valuable
assistance to the users of a transportation transactions
management system, namely FTMarket (Freight
Transportation Market). FTMarket is fully implemented
and handles various types of transportation transactions
[14, 10]. It exploits a series of dedicated software agents
that represent and act for any type of user involved in a
transportation scenario (such as customers who look for
efficient ways to ship their products and transport
companies that may - fully or partially - carry out such
requests), while they cooperate and get the related
information in real-time mode [24]. Our overall approach
is based on flexible models that achieve efficient
communication among all parties involved, coordinate the
overall process, construct possible alternative solutions
and perform the required decision-making [10, 12]. In
addition, FTMarket is able to handle the complexity that is
inherent in such environments [6], which concerns
freighting and fleet scheduling processes, as well as
“modular transportation solutions” 1 . FTMarket provides
the customer with a set of alternative solutions for each
requested transaction. These solutions are constructed
through the use of a specially developed algorithm for
retrieving optimal and sub-optimal solutions. Moreover,
through a dedicated recommender agent [9, 22], which
builds on Web Services concepts [26], the system assists
the customer further towards making the appropriate
decisions.
The remainder of this paper is structured as follows:
Section 2 reports on background issues from the area of
recommender systems, paying particular attention to
recommendation approaches. Section 3 describes the basic
aspects of our approach, which concern the selection of
transportation plans and the evaluation of alternative
solutions. Section 4 focuses on issues raised during the
integration of the recommendation module, the
formulation of the recommendation policy, and the
exploitation of software agents and Web Services
technologies. Finally, Section 5 concludes the paper and
highlights future work directions.
2. Related Work
The most widely adopted recommendation techniques are
Collaborative Filtering (CF) and Knowledge Based
Recommendation (KBR), each one possessing its own
strengths and weaknesses. Collaborative Filtering (CF)
[17, 18] is the most commonly used recommendation
technique to date. The basic idea of CF-based algorithms
is to provide item recommendations or predictions, based
on the opinion of other like-minded users. In a typical CF
scenario, there is a list of m users U = {u1, u2, …, um} and
a list of n items I = {i1, i2, …, in}. Each user ui is associated
with a list of items Iui, for which the user has expressed
his/her opinion. Opinions can be explicitly given by the
user as a rating score (within a certain numerical scale), or
implicitly derived from transaction records (by analyzing
timing logs, mining web hyperlinks and so on). For a
particular user ua, the task of a collaborative filtering
algorithm is to find an item likeness that can be of two
forms:
1
To further explain this concept, consider the case where a customer
wants to convey some goods from place A to place B, while there is no
transport company acting directly between these two places. Supposing
that two available carriers X and Y have some scheduled itineraries from
A to C and from C to B, respectively, it is obvious that a possible solution
to the above customer’s request is to involve both X and Y and fragment
the intended overall itinerary to the related sub-routes. It is also noted that
these carriers may be associated with diverse transportation means, such
as trains, trucks, ships and airplanes.
•
•
25
Prediction: this is a numerical value, Pi, expressing
the predicted likeness of item i (i does not belong to
Iua) for the user. The predicted value is within the
same scale (e.g. from 1 to 5) as the opinion values
provided by ua [19].
Recommendation: this is a list of N items Ir (Ir is a
subset of I) that the user will like most (the
recommended list must contain items not already
selected by the user). This outcome of CF algorithms
is also known as Top-N recommendation [20].
On the other hand, KBR attempts to suggest objects
based on inferences about a user’s needs and preferences.
In some sense, all recommendation techniques could be
described as doing some kind of inference. Knowledgebased approaches are distinguished in that they utilize
functional knowledge; in other words, they have
knowledge about how a particular item meets a particular
user need and can therefore reason about the relationship
between a need and a possible recommendation. The user
profile can be any knowledge structure that supports this
inference. In the simplest case, as in Google, it may simply
be the query that the user has formulated. The Entrée
system and several other recent systems [23], employ
techniques from case-based reasoning for knowledgebased recommendations.
The knowledge used by a knowledge-based
recommender system can take many forms. Google uses
information about the links between web pages to infer
popularity and authoritative value [1]. Entrée uses
knowledge of cuisines to infer similarity between
restaurants. Utility-based approaches calculate a utility
value for objects to be recommended; in principle, such
calculations could be based on functional knowledge.
However, existing systems do not use such inference
mechanisms, thus requiring users to do their own mapping
between their needs and the features of products, either in
the form of preference functions for each feature, as in the
case of Tête-à-Tête, or answers to a detailed questionnaire,
as in the case of PersonaLogic [2]. Knowledge-based
recommender systems are prone to the drawback of all
knowledge-based systems: the need for knowledge
acquisition. More specifically, there are three types of
knowledge that are involved in such systems:
•
•
Catalog knowledge: Knowledge about the objects
being recommended and their features. For example,
the system should know that “Gasoline” is a type of
“Fuel”.
Functional knowledge: The system must be able to
match the user’s needs with the object that might
26
ISSN (Print): 1694-0814
•
satisfy those needs. For example, a recommendation
module should know that the transportation of toxics
require a higher safety level.
User knowledge: To provide good recommendations,
the system must have some knowledge about the user.
This might take the form of general demographic
information or specific information about the need for
which a recommendation is sought.
Of these knowledge types, the last one is the most
challenging, as it is an instance of the general usermodelling problem [25]. Despite this drawback,
knowledge-based recommendation has some beneficial
characteristics. First of all, it is appropriate for casual
exploration, because it demands less from the user
(compared to the utility-based recommendation).
Moreover, it does not involve a start-up period during
which its suggestions are of low quality. On the other
hand, a knowledge-based recommender cannot “discover”
user niches, the way collaborative systems can. However,
it can make recommendations as wide-ranging as its
knowledge base allows.
Alternative techniques have been proposed in the
literature in order to handle the above issues [11]. Having
thoroughly considered their pros and cons, our approach
follows a hybrid recommendation technique. Generally
speaking, CF and KBR techniques can be combined in
hybrid recommendation systems in order to improve their
performance. Most commonly, CF is combined with some
other technique in an attempt to minimize or avoid the
ramp-up problem [3].
3. The Proposed System
3.1 Transportation plans and evaluation of alternative
solutions
The recommendation procedure adopted in our approach
is highly associated with the selection (by the user) of the
appropriate transportation plan. A transportation plan
typically defines the user preferences for the upcoming
transactions. The five alternative plans offered are:
•
•
•
•
•
Express
Economic
Safe
Dependable
User Defined
It can be easily observed that each of the first four
plans declares a specific tension in the recommendation
strategy to be followed by the system, in that it either
minimizes the overall duration or cost (first two plans), or
it retains a high level of safety or dependability (third and
fourth plans) of the suggested itineraries. The last choice
offers the possibility for a user-customized plan definition.
Such a plan may combine parameters from all the above
four plans. The selection of one of these plans will
influence the recommendation process of our approach for
the particular user.
Figure 1: Transaction’s request interface
As shown in Figure 1, which depicts the system’s interface
for handling a user’s request, the user provides input about
the loading and delivery terminals, the quantity to be
transported, expresses his/her preferences concerning
maximum cost and duration of the transaction, and selects
the desired transportation plan. By selecting the “userdefined” plan, a new window appears, allowing the user to
adjust the criteria (cost, duration, safety, dependability) of
his/her transportation request.
Table 1: Selection criteria for the alternative transportation plans
(safety and dependability take values from the set {very low, low,
average, high, very high}).
Plan
Cost
Duration
Safety
Dependabil
ity
Express
Any
Min
Any
Any
Economic
Min
Any
Any
Any
Safe
Any
Any
>Average
≥Low
Dependable
Any
Any
≥ Low
> Average
Hybrid
User
Defined
User
Defined
User
Defined
User
Defined
27
weight (Wcost-ij) and a duration weight (Wduration-ij). It is
obvious that:
(1)
W =W
+W
ij
cost - ij
duration - ij
During the construction of the available transportation
solutions, our approach excludes solutions that do not
comply with the customer’s requirements. More
specifically, a set of predefined rules is employed to
exclude the alternative solutions that do not correspond to
the specific freight transportation’s requirements and
customer preferences. Table 1 summarizes the constraints
to be met for each transportation plan (for the “User
Defined” plan, this process takes into account the
constraints set by the user). In all cases, solutions that do
not satisfy these constraints are discarded.
3.2 A Methodology for the Selection of Alternative
Route Paths
In our former work [10, 27], we have presented an
algorithm for constructing optimal (direct or modular)
solutions for a requested transportation transaction. This
algorithm was taking into account the cost and duration of
each sub-route, as well as the cost and duration upper
bounds (as they had been set by the user). If no optimal
solution could be constructed, the algorithm terminated
without providing any solutions. To better handle such
cases, our approach uses an elaborated version of
Dijkstra’s shortest path algorithm [4] to construct suboptimal solutions. Even if such solutions cannot be
characterized as optimal, they represent acceptable
alternatives for a specific transportation request.
As it can be retrieved from the related literature [4],
shortest path algorithms use a bidirectional, singleweighted graph to represent a connected set of vertices
(Vi) through a number of arcs Aij (from Vi to Vj). Our
algorithm takes into consideration each Aij and its
correspondent weight (Wij) in order to produce a route
path from a starting point (S) to an ending point (E) that
minimizes the total weight (WSE). The complexity of our
approach consists in the presence of a pair of variables
that affect each arc’s weight, namely the cost and the
duration. Due to the fact that there exist two weights for
each arc (cost and duration), we confronted the problem of
unifying these weights into a single one, in order to
proceed with the ranking of the solutions. As shown in
Figure 2, each arc’s Aij weight (Wij) consists of a cost
Figure 2: A hypothetical 2-weighted graph.
Having defined the total weight for each arc (Aij), we
encountered the problem of adding these two parameters
that are measured in different units (Euros and hours,
respectively). This problem was confronted by applying a
normalization technique that divided both the costij and
durationij of an arc with its correspondent maximum cost
and duration of the sub-route. It is:
W duration - ij =
Wcost -ij =
duration ij
(2)
max(duration ij )
cost ij
max(cost ij )
(3)
Another issue that came up after the weight normalization
procedure concerned the solutions’ ranking. To address
this problem, our approach provides the user with different
solutions by using a pair of weight coefficients
(costCoef and durationCoef) and by calculating
solutions corresponding to alternative combinations of the
weights of the cost and duration criteria (see Figure 3),
according to the formula:
Wij = (costCoef * Wcost - ij ) + (durationCoef * Wduration - ij ) (4)
The cost and duration coefficients take values from the set
{0, 0.1, 0,2, …, 1}. The main idea of this process is to
provide the algorithm with alternative weights (wij), each
one expressing a different combination of cost and
28
ISSN (Print): 1694-0814
Coefficients'
Significance
duration parameters. At the beginning of this procedure,
we calculate the weight of each sub-route by taking into
consideration only the duration parameter (we set the cost
coefficient to 0 and the duration coefficient to 1). Then, in
a step-wise way, we decrease the duration coefficient by
0.1 (obviously, we increase at the same time the cost
coefficient by 0.1). Finally, we calculate the sub-route’s
weight taking into consideration only the cost parameter
(the duration coefficient has become 0).
ranked by default according to the cost; in any case, users
may request alternative rankings by clicking on the
corresponding column header.
100%
90%
80%
70%
60%
50%
40%
30%
20%
10%
0%
Figure 4: Solutions produced by the system.
1
2
3
4
5
6
7
8
9
10
11
4. Integrating a Recommendation Module
Number of Iteration
Duration Coef
Cost Coef
Figure 3: Weight coefficients’ variation.
This process is described in pseudo-code as follows:
{
costCoef Å 0.0;
durationCoef Å 1.0;
step Å 0.0;
while step ≤ 1.0 calculate
{
costCoef Å step;
durationCoef Å 1-step;
weight[i][j] Å costCoef*Wcost +
durationCoef*Wduration;
perform shortest path algorithm;
step Å step + 0.1;
}
}
The outcome of the above process is then presented to
user. As shown in Figure 4 (which depicts an instance of
the related system interface), the optimal routes for a
transportation request from Athens to Patra have been
retrieved (after a related request). The basic characteristics
of each route are presented in the main table of the web
interface. By selecting the “View Details” option, the user
is able to receive an analytical description of the subroutes contained in each itinerary, as well as their
corresponding characteristics. Solutions at this phase are
4.1 A Hybrid Recommendation Methodology
The recommendation procedure begins immediately after
the abovementioned construction of the alternative
solutions. It is a complex process which is carried out in
three basic phases, which are:
•
the evaluation of the carriers and the transactions
data;
• the exploitation of transaction data through a data
mining process, and
• the recommendation methodology selection or
synthesis.
At the beginning of the process, the system stores all
the appropriate data that are submitted by the user and are
related with pending or completed transportation
transactions. These data are of significant importance and
will be further exploited by the data mining process.
Moreover, in this phase the user evaluates (i.e. assigns a
score to) the carrier(s) involved in a transaction through an
appropriate interface.
The second phase of recommendation concerns the
data mining process. Data mining is a useful decision
support technique, which can be used to find trends and
regularities in big volumes of data. At this phase,
transactions data are gathered through knowledge
construction processes. In our case, the data mining
process constructs a model from the recommendation
module’s database that may produce well defined
knowledge rules. This procedure is performed through
SQL queries performed on the transactions’ tables. After
the completion of this process, the constructed knowledgebased rules participate in the production of knowledgebased recommendation data that will be evaluated and
synthesized in the last phase of recommendation.
The last phase of recommendation refers to the
selection or synthesis of the appropriate recommendation
technique. This objective will be reached through the
definition of well structured rules that will be applied for
each transaction. The Recommender Agent of our system
takes the initiative to select the most appropriate
recommendation technique. For example, for a particular
itinerary from point i to point j, taking into consideration
that the customer has selected a certain plan, a rule for the
specific itinerary could lead to the recommendation of a
carrier that is different than the one suggested by the CF
technique, based on the carriers’ evaluation process
described earlier in this section. The recommendation
methodology described above is graphically presented in
Figure 5, through a data flow diagram.
29
Table 2: Recommendation Module’s Database Model
Table Name
Description
Transactions
Transactions in progress
Transaction’s Subroutes
Transactions sub-routes in progress
Transactions _Rating
Completed Transactions’ evaluation
Carriers_Rating
Carrier evaluation with completed
Users_Reliability
Customers reliability evaluation
Temp_Transactions
Proposed transaction itineraries
Temp_Transactions_Subroutes
Subroutes of the proposed itineraries
4.2 Calculation of Recommendation Score
Figure 5: The data flow diagram for the recommendation methodology
Due to the large amount of data the recommendation
module takes into account in order to provide knowledgebased recommendations, the database model has been
thoroughly considered. The system’s database has been
designed through the use of SQL Server 2005
Management Console, in order to accomplish further with
the customers’ needs. Much attention has been paid into
the reorganization of data tables’ fields, as well as into the
representation of the entities’ relationships [16]. The
database model that participates in the knowledge
construction of the recommendation’s phase is presented
in Table 2.
After the ranking phase, the evaluation of each alternative
route retrieved is performed. Our system retrieves all
possible transportation routes that can be constructed for a
given transaction request. These routes are presented to
the user through an appropriate designed user interface.
The corresponding user interface enables the user to either
select one of the proposed routes (in this case, he/she will
be asked to complete the transaction), or to be redirected
to a user-friendly interface where he/she can receive
recommendations for each separate route. The evaluation
of a transaction is based on various criteria, such as:
•
•
•
•
•
•
•
•
Cost
Duration
Safety
Reliability
Average scores of the above carriers’ elements.
Average scores of the sub-routes contained in the
transaction
The number of times that the specific route has been
selected by other customers (popularity).
Number of transloadings
30
ISSN (Print): 1694-0814
The recommendation procedure is implemented through
the evaluation of both the transactions and the
transportation companies involved. It is a complex
procedure, basically due to the fact that a modular solution
may involve two or more carriers. It is obvious that a
transaction can receive an overall negative evaluation,
while - at the same time - a specific part could have been
completed quite satisfactorily. The evaluation of a
transaction is based on a set of criteria such as cost,
duration, safety, dependability, average score of a carrier,
itinerary’s popularity and number of transloadings [15].
Taking into consideration all the above issues, we define
total
the calculation formula of the overall score Oi, j of each
transaction from point i to point j (for each sub-route of
the itinerary). It is:
(
)
Oi,total
= Oi,t j + Oi,s j + Oi,r j
j
(5)
⎡avg(Ci,r j * ur) + avg(Tr * ur) ⎤⎦ c
Oi,r j = ⎣
2
(10)
where
Ci,t j = The carrier’s score according to time, for the
transportation from point i to j.
C si, j = The carrier’s score according to safety, for the
transportation from point i to j.
Ci,r j = The carrier’s score according to dependability, for
the transportation from point i to j.
Tt = The transaction’s score according to time.
Ts = The transaction’s score according to safety.
Tr = The transaction’s score according to dependability.
The expression avg(x) refers to the average value of the
element x in the database, and the variables a,b,c are
O i, j =
final
n
2
(O i,total
- Ο i,cost
j
j )
i, j = 1
f S, E
∑
coefficients related with the user’s preferences according
(6)
t
s
r
where Oi , j , Oi , j , Oi , j represent the score of the time,
safety and dependability, respectively, for the
transportation from point i to point j . The variable
fS,E represents the number of transloadings of each
proposed solution and is considered as a negative factor,
assuming that a large number of transloadings could evoke
damage in the product and increase the transaction’s
completion time. The number of transloadings is related to
the number of sub-routes (n) of each itinerary. It is:
fS, E = n - 1, n > 1
(7)
Each one of the detailed scores is calculated according to
the score that has been assigned to the carrier and each
sub-route. It is:
⎡ avg(Ci,t j * ur) + avg(Tt * ur) ⎤⎦ a
(8)
Oi,t j = ⎣
2
⎡ avg(Ci, j * ur) + avg(Ts * ur) ⎤⎦ b
Osi, j = ⎣
s
2
(9)
to time, safety and dependability respectively. Having
defined the detailed scores for each sub-route, we
calculate the overall score
(O )
total
S, E
for the proposed
itinerary from point S (start) to point E (end).
O
to ta l
S,E
n
=
∑
i, j = 1
⎧ O i,t j + O i,s j + O i,r j ⎫
⎨
⎬ (11)
⎩ (a + b + c ) * n ⎭
For the calculation of
( O ) we
total
S,E
do not take into
consideration the proposed cost of a transaction, due to the
fact that the system evaluates it through its normalization.
The evaluation of the cost is performed through the
formula:
Oi,cost
j =
(
cost i, j
(
min cost i, j
)
(12)
)
where min cost i, j represents the minimum cost for the
specific route. At this point we encapsulate into the overall
score the cost’s score in order to recalculate a final score
( O ) for
final
i, j
the transaction, which will be the system’s
final recommendation to the user. It is:
⎡ (O i,tojta l - Ο i,c ojst ) 2 ⎤
⎥ (13)
f
i, j = 1 ⎢
⎥⎦
S
,
E
⎣
n
O i,finj a l = ∑ ⎢
4.3 An Example
This subsection presents an example of the
recommendation process and its runtime environment.
Having performed the optimal routes retrieval algorithm
[4, 15], the user is transferred to the recommendation
interface, where the results of the recommendation process
are presented (Figure 6). At this phase, the evaluation of
the itineraries is executed. More specifically, for every
solution that has been retrieved for a requested transaction,
the user may further consider its sub-routes. For each subroute, the system calculates the average score that the
carrier has received for its reliability during the
transaction, as well as the average score for the
transaction’s duration. During the calculation of the above
averages, the scores that each carrier (or each route) has
received are multiplied by a user’s reliability coefficient.
This is performed in order to add a level of significance
into a reliable user’s opinion (compared with a less
reliable one). Reliability refers to the number of times that
a user has rated an itinerary, and not by the fact that
his/her evaluation was considered as being strict or not. In
addition to the above evaluation, a similar procedure takes
place with respect to the safety and the overall carrier’s
reliability during the transaction. Both the average score of
the specific elements (duration, reliability, safety, general
reliability) and the overall score are stored in the system’s
database. When this procedure is completed for all
itineraries’ sub-routes, an average of all scores is
extracted. The final score of the itinerary is the sum of the
carriers’ and the sub-routes’ overall score, normalized by
the overall cost and the number of intermediate
transloadings. Moreover, the system retrieves information
related to the completion of the above itineraries and their
correspondent frequency. This procedure aims at checking
whether a specific itinerary is constantly selected by other
users. The popularity of each route is presented to the user
later, in order not to affect his/her decision.
Initially, the recommended solutions are shown to the
user according to their final score (top table of the
interface shown in Figure 6). The user may then see each
solution’s details; by clicking on the “View Details” link
(which appears at each entry of the top table), the interface
expands dynamically and a second table appears (entitled
31
“Sub-Route Details”), containing information about the
sub-routes of the selected itinerary and the overall scores
of each sub-route. Clicking on the “More Details” link,
the user is provided with additional information about
each sub-route (such as scores for its duration, safety and
reliability). Moreover (by exploiting the “Show” link at
the “Top-10 Carriers” column), the user is given the
opportunity to compare a sub-route’s carrier with any of
the Top-10 carriers that exist for the particular sub-route
(this is a common practice in CF techniques). In such a
case, the interface of Figure 6 expands further and a third
table, entitled “Top-10 Carriers”, appears. When
selecting a carrier from this table, by clicking on the
“Select” link, the corresponding differences (in terms of
cost, duration and carrier’s rating) are presented in the
bottom right part of the window (under the header
“Additional Features”).
Figure 6: The recommendation module interface.
4.4 Implementation Issues
A new software agent, namely the Recommender Agent
(RA), has been implemented and interconnected with a
correspondent Web Service, in order to coordinate the
overall recommendation process. The main tasks of the
RA concern the coordination of the recommendation
module, depending on the characteristics of each
transaction. Through these formally modeled tasks, RA
provides continuous assistance to customers, while it
remains active and capable to adapt its “behavior” into a
rapidly changing environment. RA is responsible for the
coordination of the whole process, as it interacts with the
other software agents of the system [10]. Moreover, the
32
ISSN (Print): 1694-0814
recommendation policy of our system builds on Web
Services concepts [26]. A Web Service is a URLaddressable software resource that performs functions and
provides answers. It is constructed by taking a set of
software functionality and wrapping it up so that the
services it performs are visible and accessible to other
software applications. A Web Service can be discovered
and leveraged by other Web Services, applications, clients,
or agents. In other words, Web Services can request
services from other Web Services, and they can expect to
receive the results or responses from those requests.
Moreover, Web Services communicate using an easy-toimplement standard protocol (SOAP). Web Services may
interoperate in a loosely-coupled manner; they can request
services across Internet and wait for a response [5]. Due to
the fact that external applications could exploit the
proposed recommendation services, the implementation of
the FTMarket’s recommendation module was performed
according to Web Services concepts and standards.
SOAP envelope. This protocol can work across a variety
of mechanisms, either asynchronously or synchronously.
Web Services may make requests of multiple services in
parallel and wait for their responses. The set of services to
be provided in the FTMarket platform will be increased in
the future (it will constitute a services repository). It is
noted that it is not necessary for all these services to be
provided through a single server; multiple servers, located
in distinct providers, may be used. Finally, our system’s
Web Services are message-based. Interaction via message
exchange means that instead of a client invoking
functionality exposed as a Web Service, it sends a request
to the Web Service to have the functionality invoked [7,
8]. In other words, what a Web Service exposes is the
functionality of receiving a message. We have adopted a
generic message interchange, which means that delivery of
message content is independent of its format.
5. Conclusions
Figure 7: The recommendation module architecture.
The overall architecture of the FTMarket’s
recommendation module is illustrated in Figure 7. As
shown, the module is appropriately wrapped in order to
describe the kind of service to be provided. To be easily
located by users, such descriptions of services are placed
in a shared public registry. It is through this registry that
users may look up for the services they need each time (in
any case, a Web Service can be directly accessed if one
knows its URL and WSDL). The correspondent agent that
needs functions provided by the specific Web Service
sends the appropriate request as an XML document in a
This paper has elaborated a series of issues related to the
integration of hybrid recommendation techniques into an
agent–based transportation transactions management
platform. We proposed a hybrid recommendation module
that combines different recommendation techniques in
order to provide the user with more accurate and efficient
suggestions. The overall recommendation process is
coordinated by a software agent, which is responsible for
carrying out multiple tasks, such as coordination of the
recommendation module, selection of alternatives and
knowledge synthesis through the exploitation of different
recommendation techniques and algorithms. The presence
of the Recommender Agent guarantees that the user will be
provided with continuous recommendations, which are
dynamically updated. Finally, we have exploited concepts
related to Web Services in order to make the proposed
recommendation functionalities accessible from external
applications.
Future work plans mainly concern the consideration of
additional recommendation techniques, such as content–
based or model–based techniques and the exploitation of
data mining algorithms in order to enhance the overall
quality of the recommendations provided. The
development of additional (local or remote) Web Services,
which will be capable of carrying out more complex
requests for recommendation techniques synthesis, is
another major concern.
References
[1] S. Brin, L. Page, The anatomy of a large-scale hyper textual
Web search engine, Computer Networks and ISDN Systems
30(1-7) (1998) , pp. 107-117.
[2] R. Burke, Hybrid Recommender Systems: Survey and
Experiments, User Modelling and User-Adapted Interaction
12 (2002), pp. 331-370.
[3] R. Burke, Integrating Knowledge-Based and CollaborativeFiltering Recommender Systems, Artificial Intelligence for
Electronic Commerce, AAAI Technical Report WS-99-01
(1999), pp. 69-72.
[4] A. Crauser, K. Mehlhorn, U. Meyer, P. Sanders, A
Parallelization of Dijkstra's Shortest Path Algorithm,
Proceedings of 23rd International Symposium, MFCS'98
(1998), Brno, Czech Republic.
[5] G. Fox, W. Wu, A. Uyar, H. Bulut, A Web services
framework for collaboration and audio videoconferencing,
International Multiconference in Computer Science and
Computer Engineering, Internet Computing (2002), Las
Vegas, USA.
[6] G. Froehlich, H. J. Hoover, W. Liew, P. Sorenson,
Application Framework Issues when Evolving Business
Applications for Electronic Commerce, Information Systems
24(6) (1999), pp. 457-473.
[7] D. Greenwood, M. Calisti, Engineering Web Service - Agent
Integration, IEEE Conference of Systems, Man and
Cybernetics (2004), The Hague.
[8] E. Hanson, P. Nandi, D. Levine, Conversation Enabled Web
Services for Agents and e-Business, Proceedings of
International Conference of Internet Computing, Computer
Science Research, Education and Applications (CSREA)
Press (2002), pp. 791-796.
[9] N. Jennings, P. Faratin, T.J. Norman, P. O'Brien, B. Odgers,
Autonomous Agents for Business Process Management,
International Journal of Applied Artificial Intelligence 14(2)
(2000), pp. 145-189.
[10] N. Karacapilidis, A. Lazanas, G. Megalokonomos, P.
Moraitis, On the Development of a Web-based System for
Transportation Services, Information Sciences, 176(13)
(2006), pp. 1801-1828.
[11] N. Karacapilidis, L. Hatzieleftheriou, A hybrid framework
for similarity-based recommendations, Business Intelligence
and Data Mining, 1(1) (2005), pp. 107-121.
[12] N. Karacapilidis, P. Moraitis, Building an Agent-Mediated
Electronic Commerce System with Decision Analysis
Features, Decision Support Systems 32(1) (2001), pp. 53-69.
[13] M. Klusch, K. Sycara, Brokering and Matchmaking for
Coordination of Agent Societies: A Survey, A. Omicini et al.
(eds.), Coordination of Internet Agents, Springer (2001), pp.
197-224.
[14] A. Lazanas, C Evangelou N. Karacapilidis, Ontology-Driven
Decision
Making
in
Transportation
Transactions
Management, Witold Abramowicz (ed.), Proceedings of the
8th International Conference on Business Information
Systems (2005), Poznan, Poland, pp. 228-241.
33
[15] A. Lazanas, N. Karacapilidis Y. Pirovolakis, Providing
Recommendations in an Agent-Based Transportation
Transactions Management Platform, Proceedings of the 8th
International Conference on Enterprise Information Systems
(2006), Paphos, Cyprus.
[16] U. Nahm, R. Mooney, Text Mining with Information
Extraction, Proceedings of the AAAI Spring 2002
Symposium on Mining Answers from Texts and Knowledge
Bases (2002), Stanford, pp. 60-68.
[17] M. O’Mahony, N. Hurley, C. Silvestre, Promoting
Recommendations: An attack on Collaborative Filtering,
Proceedings of DEXA 2002 Conference, Springer-Verlag
(2002), Berlin, pp. 494-503.
[18] P. Resnick, N. Iacovou, M. Suchak, P. Bergstrom, J. Riedl,
GroupLens: An Open Architecture for Collaborative
Filtering of Netnews. In Proceedings of CSCW (1994),
Chapel Hill, NC.
[19] B.M. Sarwar, G. Karypis, J.A. Konstan, J. Riedl, Analysis of
Recommender Algorithms for E-Commerce, Proceedings of
the ACM Conference on e-Commerce, Minneapolis, ACM
Press (2000), pp. 158-167.
[20] B.M. Sarwar, G. Karypis, J.A. Konstan, J. Riedl, Item-Based
Collaborative Filtering Recommendation Algorithms,
Proceedings of the 10th International World Wide Web
Conference (2001), Hong Kong, ACM Press.
[21] J.B. Schafer, J. Konstan, J. Riedl, Electronic Commerce
Recommender Applications, Journal of Data Mining and
Knowledge Discovery, 5 (1-2) (2000), pp. 115-152.
[22] W. Shen, D.H. Norrie, An Agent-Based Approach for
Dynamic Manufacturing Scheduling, Proceedings of
Autonomous Agents'98 Workshop on Agent-Based
Manufacturing, Minneapolis (1998), pp. 117-128.
[23] S. Schmitt, R. Bergmann, Applying case-based reasoning
technology for product selection and customization in
electronic commerce environments, 12th Bled Electronic
Commerce Conference (1999), Bled, Slovenia.
[24] K. Sycara, D. Zeng, Coordination of Multiple Intelligent
Software Agents, International Journal of Cooperative
Information Systems, 5(2-3) (1996), pp 546-563.
[25] B. Towle, C. Quinn, Knowledge Based Recommender
Systems Using Explicit User Models, Knowledge-Based
Electronic Markets, AAAI Technical Report WS-00-04,
AAAI Press (2000), pp. 74 -77.
[26] H. Wang, J. Huang, Y. Qu, J. Xie, Web services: problems
and future directions, Journal of Web Semantics, 1 (2004),
pp. 309–320.
[27] A. Lazanas, G. Megalokonomos, Optimizing Alternative
Routes Retrieval in an Agent–based Transportation
Management System. Proceedings of the International
Conference on Service Systems and Service Management
(ICSSSM 2006), Troyes, France, pp. 1525-1530.
Dr. Alexis Lazanas studied Applied Informatics in Athens
University of Economic and Business (B.Sc. 1996) and received
his Ph.D. from University of Patras (Greece) in the field of
Recommender
Systems,
Data
Mining
and
Intermodal
Transportation (2008). He worked in Technological Educational
Institute (T.E.I.) of Patras as Scientific Collaborator and as
ISSN (Print): 1694-0814
Software Developer – Special Analyst in various major companies.
Currently he is working as Teacher of Informatics in Greek Public
Education. His research interests are on the areas of Agent-based
Information Systems, Data Mining, Web Technologies, Hybrid
Recommender
Systems
and
Intermodal
Transportation
Management.
34
35
ISSN (Print): 1694-0814
Geometric and Signal Strength Dilution of Precision (DoP)
Wi-Fi
Soumaya ZIRARI*, Philippe CANALDA and François SPIES
Computer Science Laboratory of the University of Franche-Comté, France
Numerica, 1 cours, Louis Leprince Ringuet, 25200 Montbéliard
Abstract
The democratization of wireless networks combined to the
emergence of mobile devices increasingly autonomous and
efficient lead to new services. Positioning services become
overcrowded. Accuracy is the main quality criteria in
positioning. But to better appreciate this one a coefficient is
needed. In this paper we present Geometric and Signal
Strength Dilution of Precision (DOP) for positioning systems
based on Wi-Fi and Signal Strength measurements.
Keywords: Wireless LAN, Radio position measurement,
Indoor radio communication.
The GPS is limited in given environments and Wi-Fi is
becoming a viable positioning method. The authors
think that the Wi-Fi network can be adapted by learning
from the GPS.
In this paper, we present a mathematical approach of a
new version of the known GPS Dilution of Precision [6]
which is more adapted to the Wi-FI networks and use
other elements to estimate the precision. We also present
a model that allows to estimate the precision based on
criteria other than the geometric one only.
The third section presents and analyzes some results.
1. Introduction
2. GEOMETRIC CRITERIA
The world population is currently growing which
implies a remarkable increase in buildings and
skyscrapers. These are obstacles for Global Navigation
Satellite Systems (GNSS) such as the Global Positioning
System (GPS). New networks have emerged (UMTS,
GSM, ...) which does not help to reduce the impact of
interferences. These factors among others contribute to
the GPS [1] up to 20 meter loss in accuracy especially
in urban and peri-urban environments.
During the last ten years the number of users of the
IEEE 802.11x community has known a remarkable
growth and a new positioning solution based on Wi-Fi
was born. Some positioning algorithms guaranty an
accuracy of 5 meters such as RADAR[2], Viterbi-like
algorithm [3], Friis and Reference Based Hybrid Model
[4] (FRBHM) [5].
The evolution of the IEEE 802.11 standard fulfil more
and more the constraints allowing the improvement of
its efficiency in large and more complex environments.
The efficiency of such networks is measured by different
criteria. Some of those criteria are focused on the
network geometry, others on the throughput [7] or on
the interference [8].
A. Gondran and al. [9] provide a geometric indicator for
WLAN planning. This indicator is based on the study of
the covered area by a Basic Service Set (BSS), where a
cell relative to one antenna is a set of pixels associated
to a given base station. The cell C is defined by:
c= {bi , j / F i , j q }
(1)
IJCSI
44
Whereis bi , j the pixel of coordinates i , j and F i , j is
the signal strength received at bi , j exceeding a given
quality threshold q.
Considering the 2-D space, each pixels have 8
neighbours with the exception of the pixels on space
borders. Mabed and al. [10] define the geometrical
criteria as bellow:
3. Propagation Models
When the signal transmitted by a transmitter travels in
space, it loses its power. Part of the energy of the signal
strength is dissipated. The environment where the
carrier signal travels and the distance covered have an
important impact on the signal attenuation.
Several equations have been developed.
(2)
3.1 FRIIS
A. Gondran and al. adapted this formula to 3-D space
which can be indoor environment such as buildings.
The Friis [11] equation is:
(5)
where :
(3)
•
P R and P T are respectively the Signal Strength
(SS) received and the SS emitted;
Where k presents the floor.
•
The geometric indicator regrouping all floor-indicators
is defined by the following equation:
transmitter antenna gains;
•
•
(4
G R and G T are respectively the receiver and
is the carrier wavelength;
d is the distance between the receiver and the
transmitter.
)
Where
3.2 Interlink Networks
The Interlink Networks [14] approach offers to replace
the power 2 in the Friis formula by the power to the 3.5
due to the prompt wave's attenuation in a building
because of the high number of obstacles in this one.
The Interlink Networks formula is:
(6)
where :
IJCSI
•
43
P R and P T are respectively the Signal Strength
(SS) received and the SS emitted;
•
G R and G T are respectively the receiver and
transmitter antenna gains;
The
is the carrier wavelength;
•
d is the distance between the receiver and the
•
2. Contribution
transmitter.
contribution
of
this
paper
consists
in
presenting a precision of dilution model for wireless
networks. This model aims at giving an idea about
the position estimation accuracy. This model can
be described in three steps:
3.3 SNAP-WPS
Y. Wang proves in the paper [15] the possibility to
approximate the target position by measuring the signal
strength. In fact, the signal attenuation between the
transmitter and the receiver allows to determine the
mobile position. However, the Friis equation enables to
estimate the distance between the receiver and the
transmitter in an environment without any obstacles.
Thus, Y. Wang suggest an empirical model based on
regression. By comparing the residual among different
degrees polynomials, he decide that a cubic regressive
equation would be adequate for the empirical
2
model EM :
d i = 0.000198 S 3i − 0.025 S 2i
1.14 S i − 14.8 (
1- The first step consists in the constitution of a set
of all visible access points (Fig. 1). The number of
visible access points is one of the decisive
elements on the accuracy of a positioning system.
Our needs in the number of visible access points
depend on the dimension of the positioning system.
At least three APs for a two dimension positioning
system and at least four APs for a three dimension
one. If the number of AP is not sufficient, we set
automatically the value of the precision of dilution
coefficient as infinite. The optimal value is equal to
7)
Where S is the signal strength (SS) in dBM, normally is
between 15-90 dBM.
one.
2- The second step concerns the signal strength of
the visible access points (Fig. 2). We assume that
access points with a signal strength under a given
3.4 Analysis
threshold may induce errors in the position
The results in Table.1 [16] present the comparison
between Wi-Fi positioning systems.
Table 1: Comparison Between The Positioning Algorithms
estimation of the target. An access point with a bad
signal strength can be near or far from the user. In
fact, the signal strength may be attenuated either
Positioning
System
Mean
Error
Standard
Deviation
Friis
9.86
6.3
SNAP-WPS
8.76
5.87
Interlink
Networks
9.58
5.11
FBCM
7.77
3.03
3- The final step deals with the positioning system
Radar
4.62
2.98
architecture geometry i.e. the third step verifies if
FRBHM
5.98
3.22
the visible access points are geometrically well
because of the distance or because of the number
of obstacles. If only three access points have a
good signal strength (we are in a 3D positioning
system) we predict that the coefficient value will be
higher.
IJCSI
distributed with respect to the user. For this step
we propose a Wi-Fi DOP Dilution Of Precision
which is calculated as below.
44
X c ,i − X u
d i=
2
Y c , i− Y u
2
Z c ,i − Z u
2
(8)
X
c ,i
,Y
c ,i
,Z
c ,i
are
the AP i coordinates
and X u , Y u , Z u the user unknown coordinates.
We obtain:
(9)
4.1 Friis equation
The Friis equation [13] as seen before is:
The Friis equation allows us to compute the distance as
below:
d i=
PT ,i G RGT ,i
4
P R ,i
(10)
Where :
Let us suppose S AP = N AP the number of visible
access points. We assume that :
P R , i , P T , i ,G R and G T , i are respectively the receiver
and AP i data.
S AP = {AP 1 , AP 2 , ... , AP N AP }
Where AP i are the visible access points.
The radius of circle d i ( i
{1,... , N } the number
of calculation) is defined by:
AP
The distance d i can be approximated by a Taylor
expansion:
(11)
The Taylor expansion at the first order is:
IJCSI
43
(12)
and :
Where
X c ,i − X u
bi , x =
, bi , y =
ri
,
ri
We obtain:
Z c , i− Z u
b i , z=
ri=
Y c ,i − Y u
X c , i− X u
C
and
ri
2
Y c , i− Y u
2
Z c ,i− Z u
2
P R= H
X (16)
Where C is a known matrix equal to:
We obtain:
(13)
and ci =
P T ,i G T , i G R
4
The linear system is:
d= H
X (14)
Where :
We
that P T , i ,G R and G T , i are
suppose
P R , i=
fixed
1
P R,i
−
1
P R,i
parameters. Only P R , i the Signal Strength (SS) received
The G matrix is defined by:
G= H T H
from the AP i is unknown and then estimated.
Thus from the equation (5), we obtain:
d i=
We have:
PT , i G T , i G R
4
−1
(17)
The Wi-Fi GDOP follows the equation bellow:
1
P R,i
−
1
P R,i
(15)
DOP = Tr [ G ] (18)
We conclude from the model that we can estimate the
positioning accuracy, and measure the error of the
IJCSI
wireless positioning system by analysing the following
elements:
44
Where :
• P R Signal Strength (SS) received from the AP which
emits a SS P T
• G R the user antenna gain and G T the AP antenna
gain;
• the carrier wavelength;
• The number of visible AP.
The G matrix is defined by:
G= H T H
−1
The Wi-Fi GDOP follows the equation bellow:
4.2 Interlink Networks
DOP = Tr [ G ]
The Interlink Networks formula is:
4.3 SNAP-WPS
The distance in SNAP-WPS system is equal to:
The distance is:
d i = 0.000198 S 3i − 0.025 S 2i
1.14 S i − 14.8
The linear system is:
S= H
(19)
The linear system become equivalent to:
C P R= H X
Where C is a known matrix equal to:
X (20)
5. Experiments
Experiments have been carried out to validate our
model of precision dilution for wireless networks. Open
Wireless Positioning System (OWLPS) [17], which is
an indoor positioning system, based on the Wi-Fi
wireless network, was the positioning system used to
calculate the mobile position. The experiments were
carried in our laboratory, Laboratoire d' Informatique
de Franche Comté (LIFC).
5.1 OWLPS Architecture
and
IJCSI
43
Fig. 3. : The environment of experimentation
5.3 Analysis
Open Wireless Positioning System (OWLPS)
implements several positioning techniques and
algorithms such as FBCM [4] or FRBHM [5]. The
system is Infrastructure-centred, i.e., the mobile asks its
position to the infrastructure (see Fig. 3). The main task
of the system is to provide an adequate environment to
the creation and test of new techniques, propagation
models and for the development of hybrid techniques
combining existing algorithms.
The first experiments were done in order to verify the
impact of the number of access points on our model and
to check if the Wi-Fi DOP is consistent with this
information.
5.2 The experimentation scenario
As we can see in Fig.4, the experimentation scenario
was about a mobile displacements during an interval of
time. During all this interval, the user is located through
the OWLPS system and the algorithm used for the
positioning are Friis, Interlink Networks and FRBHM.
Along all the mobile trajectory, we know the exact
mobile coordinates and the estimated one, which allow
us to analyse the results. The positioning system is a 3D
one.
Fig. 6. : The DOP cartography when the mobile is moving in the first floor
The Fig. 6 proves how the Geometric and Signal
Strength Dilution of Precision (DoP) Wi-Fi progress
with the mobile movement in the first floor of the
building.
IJCSI
44
Analysing the results presented in Fig.7, we deduce the
DOP fits quite well in terms of number of visible access
points. In fact, as shown in Fig.7
when the
DOP[ 10,15], the number of visible access points is
equal to three, thus the Wi-Fi DOP values reach infinite
values.
However, the Wi-Fi DOP values reach good values
when the number of visible access points is up to four
but we observe some peaks when the number acceptable
of access point for 3D positioning system is minimal
(i.e. four access points).
The second step of our experiments was done in order
to verify the impact of the signal strength of each visible
access point on our model and in which way this
information makes the DOP vary.
Fig. 8 proves that the Wi-Fi DOP is really influenced by
the signal strength of the access points. When
DOP [10,15] and DOP [35,44], the Wi-Fi DOP
values vary from seven to the infinite. If we look at the
signal strength for those behaviours we note that the
signal can not be received or the signal is too weak. This
means that the model can in fact predict the system
accuracy by analysing the access point signal strength.
IJCSI
The third step of our experiments has been carried out to
analyse the efficiency of our model by comparing the
real trajectory and the estimated one with Wi-Fi DOP
values (see Fig. 9). The analysis shows that the
trajectories (the real one and the estimated one) are
more or less similar except when the Wi-Fi DOP is up to
eight.
43
The model presented in this paper may provide the
guaranty we need. In fact, as shown in the results
obtained in the previous section, our model illustrates
the positioning system accuracy.
The idea consists in the observation of the results of the
model and when the values of this one reach a given
threshold, we inform the user that the position accuracy
is not sufficient and then anticipate a solution to
guaranty the quality and continuity of service.
7. Future Trends
Our model opens and leads to numerous extensions and
perspectives.
The coefficient of dilution of precision or rather the WiFi DOP is a good candidate to specify the most adequate
access points distribution. It is possible to extend the
Wi-Fi DOP to the system OWLPS.
It could provide a continuity of positioning, but also
assistance to the optimal positioning of access points.
The aim of this study is to offer to the user most of the
time four access points with a DOP of the order of 2 in
sight.
The fourth and last step is to verify whether the Wi-Fi
DOP is a good indicator of the positioning accuracy.
Fig.10 shows that when DOP [10,15], the error is up
to eleven. When the Wi-Fi DOP value is equal to three
(when the Wi-Fi DOP value is
[1,5], we consider that
the system has a god accuracy) the mean average error is
equal to four.
6. Conclusions
Nowadays, the Wi-Fi positioning algorithms and
systems are becoming a new mean of positioning mobile
terminals within a heterogeneous environment.
The quality of service of such system may be improved
in order to guaranty the integrity and the continuity of
service.
This paper describes a model for dilution of precision
and a mathematical description of the coefficient
weakening of the accuracy, the Wi-Fi DOP.
Acknowledgments
We thank all the reviews for their detailed feedback and
suggestions specially Matteo Cypriani.
References
[1] US Army Corps of Engineer, Engineering and Design NAVSTAR Global Positioning System Surveying,
Department of the Army, 2003, Washington, DC, July.
[2] Paramvir Bahl, Venkata N. Padmanabhan. RADAR: An
In-Building RF-Based User Location and Tracking System.
In Proceedings of the IEEE Infocom 2000, Tel-Aviv,
Israel, vol. 2, Mar. 2000, pp. 775--784.
[3] P.Bahl, A. Balachandran, V. N. Padmanabha.
Enhancements to the RADAR User Location and Tracking
System. Microsoft Research Technical Report, February
2000.
[4] F. Lassabe and O. Baala and P. Canalda and P. Chatonnay
and F. Spies, A Friis-based Calibrated Model for Wi-Fi
Terminals Positioning, Proceedings of IEEE Int. Symp. on
a World of Wireless, Mobile and Multimedia Networks
(WoWMoM 2005), 2005
[5] Frédéric Lassabe. Géolocalisation et prédiction dans les
réseaux Wi-Fi en intérieur, Rapport de thèse. 2009
IJCSI
[6] Radar, Sonar and Navigation, IEE Proceedings - Volume
147, Issue 5, Oct. 2000 Page(s):259 - 264 Yarlagadda,
R.; Ali, I.; Al-Dhahir, N.; Hershey, J., GPS GDOP metric
Radar, Sonar and Navigation, IEE Proceedings - Volume
147, Issue 5, Oct. 2000 Page(s):259 - 264 .
[7] Ling X., Yeung K.L., ?Joint access point placement and
channel assignment for 802.11 wireless LANs?, IEEE
Wireless Communication and Networking Conference, pp.
1583-1588, 2005.
[8] Amaldi E., Capone A., Cesana M., Malucelli F.,
Optimizing WLAN Radio Coverage, IEEE International
Conference on Communications 2004, 1, pp.180-184,
2004
[9] Gondran, A.; Baala, O.; Caminada, A.; Mabed, H., "3-D
BSS geometric indicator for WLAN planning" Software,
Telecommunications and Computer Networks, 2007.
SoftCOM 2007. 15th International Conference on Volume
, Issue , 27-29 Sept. 2007 Page(s):1 – 5
[10] H. Mabed, A. Caminada, Geometric criteria to improve
the interference performances of cellular network, IEEE
Vehicular Technology Conference, Montreal. Sept. 2006.
[11] S. Zirari, P. Canalda, and F. Spies. Modelling and
Emulation of an Extended GDOP For Hybrid And
Combined Positioning System. In ENC-GNSS'09,
European Navigation Conference - Global Navigation
Satellite Systems, Naples, Italy, May 2009
[12] S. Zirari, P. Canalda, and F. Spies. A Very First
Geometric Dilution Of Precision Proposal For Wireless
Access Mobile Networks. In SPACOMM'09, The First
International Conference on Advances in Satellite and
Space Communications, Colmar, France, July 2009
[13] H. T. Friis, A note on a simple transmission formula,
Proc. IRE, pp. 254-256, 1946. (NOAA), Environmental
Technology Laboratory (ETL), in Boulder, Colorado
[14] Inc Interlink Networks. A practical approach to
identifying and tracking unauthorized 802.11 cards and
access points. Technical report, 2002.
[15] Y. Wang, X. Jia, and H.K Lee. An indoors wireless
positioning system based on wireless local area network
44
infrastructure. In 6th Int. Symp. on Satellite Navigation
Technology Including Mobile Positioning and Location
Services, number paper 54, Melbourne, July 2003. CDROM proc.
[16] Matteo Cypriani, Frédéric Lassabe, Soumaya Zirari,
Philippe Canalda, François Spies. Open Wireless
Positioning System : un système de géopositionnement par
Wi-Fi en intérieur. JDIR, belfort, France, 2009.
[17] M. Cypriani, F. Lassabe, S. Zirari, P. Canalda, and F.
Spies, Open wireless positioning system, Université de
Franche-Comté, Tech. Rep. RT2008-02.
Soumaya Zirari was born in 1981. She received her diploma in
engineering in 2006. She is preparing her Ph.D Thesis at the
Computer Science Laboratory at the University of Franche-ComtŽ
in France, to be defended the 1st semester of 2010. She is
focusing on hybrid location-based services and service continuity.
Dr Philippe Canalda got M.Sc. and Ph.D. Degrees in computer
science from the University of OrlŽans (France) in 1991 and 1997,
respectively. He worked at INRIA Rocquencourt from 1991 to 1996
on the automatic generation of optimizing and parallel n-to-n crosscompilers. From 1996 to 1998, he worked as Research Engineer in
the Associated Compiler Expert start-up factory at Amsterdam, The
Netherlands. Then he worked 2 years at LORIA on the
synchronisation of cooperative process fragment, based on
workflow model, and applied to ephemeral enterprise. Since 2001,
he is an Associate Professor at the Computer Science Laboratory
(LIFC, EA 4269) at the University of Franche-Comté in France. His
research topics deal with, on the one hand mobility services and
wireless positioning, and on the other hand on robust and flexible
optimizing algorithms based on graph, automata and rewriting
theories..
Prof. François Spies received his Ph.D. and the French
“Accreditation to supervise research” Degrees in 1994 and 1999,
respectively. He was an Associate Professor at the Computer
Science Laboratory at the University of Franche-Comté in France
from 1996-1999. Since 1999, he has held a Professor position at
the University of Franche-Comté. Currently he is focusing on
managing video streams on wireless and mobile architecture.
Researches on, cooperative video cache strategies including
mobility and video quality levels, transport, congestion control and
quality of service of video streams are the main developed topics.
IJCSI
45
ISSN (Print): 1694-0814
Implementation of Rule Based Algorithm for Sandhi-Vicheda Of
Compound Hindi Words
Priyanka Gupta1 ,Vishal Goyal 2
1
M.Tech. (ICT) Student, 2Lecturer
Department of Computer Science
Punjabi University Patiala
Abstract
Sandhi means to join two or more words to coin new
word. Sandhi literally means `putting together' or
combining (of sounds), It denotes all combinatory
sound-changes effected (spontaneously) for ease of
pronunciation. Sandhi-vicheda describes [5] the process
by which one letter (whether single or cojoined) is
broken to form two words. Part of the broken letter
remains as the last letter of the first word and part of the
letter forms the first letter of the next letter. SandhiVicheda is an easy and interesting way that can give
entirely new dimension that add new way to traditional
approach to Hindi Teaching. In this paper using the
Rule based algorithm we have reported an accuracy of
60-80% depending upon the number of rules to be
implemented.
Keywords: Rule Based Algorithm, Sandhi-Vicheda,
Compound Hindi Words
I INTRODUCTION
Natural Language Processing (NLP) refers to
descriptions that attempt to make the computers
analyze, understand and generate natural languages,
enabling one to address a computer in a manner as one
is addressing a human being. Natural Language
Processing is both a modern computational technology
and a method of investigating and evaluating claims
about human language itself. It is a subfield of artificial
intelligence and computational linguistics. It studies the
problems of automated generation and understanding
of natural human languages.
A word can be defined as a sequence of
characters delimited by spaces, punctuation marks, etc.
in case of written text. A compound word (also known
as co-joined word) can be broken up into two or more
independent words. A Sandhi-Vicheda module breaks
the compound word in a sentence into constituent
words. Sandhis take place whenever there is a presence
of a swara i.e.a vowel; the presence of a consonant
with a halanta; the presence of a visarga. Sanskrit has a
well defined set of rules for Sandhi-vicheda. But Hindi
has its own rules of Sandhi-vicheda. They are,
however, not so well-defined as, and much fewer in
number than, the Sanskrit rules.
1.1 The Hindi Language
Hindi is spoken in northern and central India. Linguists
think of Hindi and Urdu as the same language, the
difference being that Hindi [5] is written in the
Devanagari script and draws much of its vocabulary
from Sanskrit, while Urdu is written in the Persian
script and draws a great deal of its vocabulary from
Persian and Arabic. More than 180 million people in
India regard Hindi as their mother tongue. Another 300
million use it as second language. Hindi is the national
language of India and is spoken by almost half a billion
people in India and throughout the world and is the
world's second most spoken language. It allows you to
communicate with a far wider variety of people in
India than English which is only spoken by around five
percent of the population. It is written in an easy to
learn phonetic script called “Devanagari” which is also
used to write Sanskrit, Marathi and Nepali. Hindi is
normally spoken using a combination of 52 sounds, ten
vowels, 40 consonants, nasalisation and a kind of
aspiration. These sounds are represented in the
Devanagari script by 52 symbols: for ten vowels, two
modifiers and 40 consonants.
II RELATED WORK
Sandhi (in linguistics) [1] is a cover term for a wide
variety of phonological processes that occur at
morpheme or word boundaries, such as the fusion of
sounds across word boundaries and the alteration of
sounds due to neighboring sounds or due to the
grammatical function of adjacent words. Internal
sandhi features the alteration of sounds within words
at morpheme boundaries, as in sympathy (syn- +
pathy). External sandhi refers to changes found at
word boundaries, such as in the pronunciation [tm
bʊks] for ten books. This is not true of all dialects of
English. The Linking R of some dialects of English is a
kind of external sandhi, as is the process called liaison
in the French language. While it may be extremely
common in speech, sandhi (especially external) is
typically ignored in spelling, as is the case in English,
with the exception of the distinction between "a" and
IJCSI
46
"an" (sandhi is, however, reflected in the writing
system of Sanskrit and Hindi). External sandhi effects
can sometimes become morphologized. Most tonal
languages have Tone sandhi, in which the tones of
words alter according to pre-determined rules. For
example: Mandarin has four tones: a high monotone, a
rising tone, a falling-rising tone, and a falling tone. In
the common greeting nǐ hǎo, both words in isolation
would normally have the falling-rising tone. However,
this is difficult to say, so the tone on nǐ is pronounced
as ní (but still written nǐ in Hanyu Pinyin).
The Sanskrit Sandhi engine software is not currently
available as a standalone application, since its local use
demands the installation of an HTTP server on the
user's host.
The Sandhi module[1] developed by RCILTSSanskrit, Japanese, Chinese at Jawaharlal Nehru
University, New Delhi. RCILTS, JNU is a resource
center for Sanskrit language of DIT, Government of
India. At JNU work started in three languages viz.,
Sanskrit, Japanese, and Chinese. Using this module the
user can get the information about Sandhi rules and
processes. Sutra number in Astyadhayi and its
description is displayed. User can learn three types of
Svara Sandhi, Vyanjan Sandhi, Hal Sandhi through this
Sandhi module Data is in Unicode. Sandhi exceptions
and options are also incorporated. This module takes
two words as input. First word cannot be null but
second word can be. A user can input the two words
and submit the form to get the result of the given input.
Chinese Tone Sandhi,[2] Cheng and Chin-Chuan
from California University, Berkeley, Phonology
Laboratory faced the problem that English stresses are
interpreted by Chinese speakers when they speak
Chinese with Engish words inserted. Chinese speakers
in the United States usually speak Chinese with Engish
words inserted. In Mandarin Chinese, a tone-sandhi
rule changes a third tone preceding another third tone
to a second tone. Using the tone-sandhi rule, they
designed the experiment to find out hoe English
stresses are interpreted in Chinese sentences. Stress
does not exist in the underlying representations of
English phonology. But in studying bilingual
phenomena, the phonetic level is also important. Fry
(1995) found that when a vowel was long and of high
intensity, listeners agreed that the vowel was strongly
stressed. The results of his experiments indicate that
the duration ratio has a stronger influence on
judgements of stress than has the intensity ratio.
Lehiste and Peterson (1959) also reported experiments
on stress.
English l-sandhi [3] involves an allophonic alternation
in alveolar contact for word-final /l/ in connected
speech [4]. EPG data for five Scottish Standard English
and five Southern Standard British English speakers
shows that there is individual and dialectal variation in
contact patterns.
III PROBLEM DEFINITION
Developing programs that understand a natural
language is a difficult task. Natural languages are large.
They contain an infinity of different sentences. No
matter how many sentences a person has heard or seen,
new ones can always be produced. Also, there is much
ambiguity in a natural language. Many words have
several meanings and sentences can have different
meanings in different contexts. Compound words are
created by joining an arbitrary number of existing
words together, and this can lead to a large increase of
the vocabulary size, and thus also to sparse data
problems. Therefore the problem of compound words
poses challenges for many NLP applications. The
problem domain, to which this paper is concerned, is
breaking up of Hindi compound words into constituent
words. In Hindi, words are a sequence of characters.
These words are combined with ‘swar’, ‘vyanjan’, and
matra’s. Hindi has its own rules of Sandhi-vicheda.
They are, however, not so well-defined as, and much
fewer in number than, the Sanskrit rules. So my
problem is to break the compound word into
constituent words with the help of rules of ‘Sandhivicheda’ in Hindi grammar. My problem is to design a
Graphical User Interface, which accepts input as a
Hindi language word (source text) from the keyboard
or mouse and break it into constituent words (target
text). The source text is converted into target text in
Unicode Format.
Compound Word
Sandhi-vicheda
ijk/khu
ij $ v/khu
HkkokFkZ
Hkko $ vFkZ
f’koky;
f’ko $ vky;
dohUnz
dfo $ bUnz
x.ks’k
x.k $ bZ’k
ijes’oj
ije $ bZ’oj
,dSd
,d $ ,d
;FkSd
;Fkk $ ,d
ijksidkj
ij $ midkj
lfU/kPNsn
lfU/k $ Nsn
foPNsn
fo $ Nsn
IJCSI
47
Table 1:Sandhi-Vicheda of Hindi Compound Words
IV IMPLEMENTATION
We have implemented the Rule-Based algorithm to
first manually find the compound words and then
develop the program that uses the database for
displaying the correct meaning to the Sandhi-Vicheda
word according to the Hindi grammar Sandhi-Vicheda
rules.
ujsUnz
uj $ bUnz
lqjUs nz
lqj $ bUnz
dohUnz
dfo $ bUnz
’kphUnz
’kph $ bUnz
Table 4: Rule III Implemented Word List
4.1 Algorithm
Step 4.4: (Rule for “Sign-E( h )” replaced with Swar
hword = hindi word to be entered
cur = Variable that stores the length of string
Step 1: Repeat for every word of the input string.
Step 2: Count the Length of String.
Step 2.1: Store the Length of String in variable.
For i = 1 To Len(hword)
cur = Mid$(hword, i, 1)
Step 3: Find the position of Matra.
hword.Substring(b - 1, 1)
Step 4: Apply the rules for sandhi –vicheda
Step 4.1: (Rule for “Sign-AA( k )” replaced with Swar
“Letter-A( v )” in Sandhi vicheda)
LokFkhZ
Lo $ vFkhZ
HkkokFkZ
Hkko $ vFkZ
lR;kFkhZ
lR; $ vFkhZ
;FkkFkZ
;Fkk $ vFkZ
“Letter-E( ई )” in Sandhi vicheda)
fxjh’k
fxfj $ bZ’k
jtuh’k
jtuh $ bZ’k
x.ks’k
ijes’oj
x.k $ bZ’k
ije $ bZ’oj
Table 5: Rule IV Implemented Word List
Step 4.5: (Rule for “Sign-U( ks )” replaced with “LetterU( m )” in Sandhi Vicheda)
ijksidkj
ij $ midkj
egksnf/k
egk $ mnf/k
vkRekaRs lxZ
vkRe $ mRlxZ
lkxjksfeZ
lkxj $ mfeZ
Table 6: Rule V Implemented Word List
Table 2: Rule I Implemented Word List
Step 4.2: (Rule for “Sign-AA( k )” replaced with Swar
“Letter-AA( vk )” in Sandhi vicheda)
Step 4.6: (Rule for “Sign-EE( S )” replaced with Vowel
“Letter-E( , )” in Sandhi Vicheda)
fo|ky;
fo|k $ vky;
lnSo
lnk $ ,o
f’koky;
f’ko $ vky;
egSo
egk $ ,o
iqLrdky;
iqLrd $ vky;
;FkSo
;Fkk $ ,o
Hkkstuky;
Hkkstu $ vky;
,dSd
,d $ ,d
Table 3: Rule II Implemented Word List
Step 4.3: (Rule for “Sign-E( h )” replaced with Swar
“Letter-E( b )” in Sandhi vicheda)
Table 7: Rule VI Implemented Word List
Step 4.7: (Rule for “Sign-EE ( S )” replaced with
“Letter-EE ( ,s )” in Sandhi Vicheda)
IJCSI
48
egS’o;Z
egk $ ,s’o;Z
nso’S o;Z
nso $ ,s’o;Z
V RESULTS AND DISCUSSION
ijeS’o;Z
ije $ ,s’o;Z
;FkSfrgkfld
;Fkk$ ,sfrgkfld
We have tested our software on more than 200 words.
Using the Rule based algorithm we have reported an
accuracy of 60-80% depending upon the number of
rules to be implemented. SANDHI-VICHEDA is an
easy and interesting way that can give entirely new
dimension that add new way to traditional approach to
Hindi Teaching.
Table 8: Rule VII Implemented Word List
Step 4.8: (Rule for eliminating the half letter in
Sandhi- Vicheda) If find the (Half CH) (PP) Letter then
eliminates the Letter and decompose the word.
lfU/kPNsn
lfU/k $ Nsn
foPNsn
fo $ Nsn
ifjPNsn
ifj $ Nsn
y{ehPNk;k
y{eh $ Nk;k
Table 9: Rule VIII Implemented Word List
Step 4.9: (Rule of Visarga in Sandhi Vicheda) If find
the (Half Letter) then replace with Sign ( : )visarga.
Total Matra=13
VI CONCLUSION AND FUTURE WORK
In this paper, we presented the technique for the
Sandhi-Vicheda of compound hindi words. Using the
Rule based algorithm we have reported an accuracy of
60-80% depending upon the number of rules to be
implemented. As future work, database can be
extended to include more entries to improve the
accuracy. This software can be used as a teaching aid
to all the students from Class-V to the highest level of
education. With this software one can learn about the
very important aspect of Hindi Grammar i.e.
‘SANDHI-VICHEDA’. By adding new more features,
we can upgrade it to learn all the aspects of Hindi
Grammar. It can also be used to solve and test the
problems related to Hindi Grammar.
fu’py
fu% $ py
fu’rst
fu% $ rst
nqLlkgl
nq% $ lkgl
ACKNOWLEDGEMENT
fuLrkj
fu% $ rkj
We would like to thank Dr. G.S. Lehal, Professor and
Head, Department of Computer Science, Punjabi
University, Patiala for many helpful suggestions and
comments.
Table 10: Rule IX Implemented Word List
Step 5: Repeat Steps 4.1 to 4.9 to check the next word
for checking the Vyanjan that combined with Matra.
Then replace the Matra with Swar.
Step 6: Find the Unicode value for each of the Hindi
characters and additional characters and use those
values to implement above rules.
Step 7: Display the results.
Our module was developed in Visual Basic.NET
(2005) and the encoding used for text was in Unicode,
most suitable for other applications as well. Unicode
uses a 16 bit encoding that provision for 65536
characters. Unicode standard [18] assigns each
character a unique numeric value and name. Presently
it provides codes for 49194 characters:
In Hindi Language:
Total Swar=13
Total Vyanjan=33
REFERENCES
[1] Bharati, Akshar, Vineet Chaitanya & Rajeev Sangal,
1991, A Computational Grammar for Indian languages
processing, Indian Linguistics Journal, pp.52, 91-103.
[2] Bharati A., Chaitanya V and Sangal R, "Natural
Language processing: A Paninian Perspective", Prentice
Hall of India, 1995.
[3] Cheng, Chin-Chuan “English Stresses and Chinese Tones
in Chinese Sentences” California University, Berkeley,
Phonology Laboratory.
[4] Dan W. Patterson “Introduction to Artificial Intelligence
and Expert Systems” Prentice Hall P-227.
IJCSI
49
[5] Elaine Rich, Kevin Knight “Artificial Intelligence” Tata
McGraw-Hill Second Edition, P-377.
[6]
Jain
Vinish
2004,
Anus¡raka:Morphological
Analyzer
Component,IIIT-Hyderabad.
Sanskrit-English
and
Dictionary
[7] James M. Scobbie (Queen Margaret University),
Marianne Pouplier (Edinburgh University), Alan A. Wrench
(Articulate Instruments Ltd.) “Conditioning Factors in
External Sandhi: An EPG Study of English /l/ Vocalisation”.
[8] Jha, Girish N., 2004, The system of Paini, Language in
India, volume4:2.
[9] Jha, Girish N. et al., 2006, Towards a Computational
analysis system for Sanskrit, Proc. of first National
symposium on Modeling and Shallow parsing of Indian
Languages at Indian Institute of Technology Bombay, pp 2534.
[10] Jurafsky Daniel and James H. Martin, 2000, Speech and
Languages Processing, Prentice-Hall, New Delhi.
[11] Kasturi Venkateswara Rao, “A Web-Based Simple
Sentence
Level GB Translator from Hindi to Sanskrit”, M.Tech(CS)
Dissertation, School of Computer Systems Sciences,
Jawaharlal Nehru University, New Delhi.
[12] Mitkov Ruslan, The Oxford Handbook of Computational
Linguistics, Oxford University Press.
[13] Peng, Shu-hui (1994). 'Effects of prosodic position and
tonal context on Taiwanese Tones'. Ohio State University
Working Papers in Linguistics, 44, 166-190.
[14] Resource Centre For Indian Language Technology
Solutions Sanskrit, Japanese, Chinese Jawaharlal Nehru
University, New Delhi “Achievements”.
[15] Scobbie, J. & Wrench, A., 2003. “An articulatory
investigation of word-final /l/ and /l/-sandhi in three dialects
of English”. Proc. XVth ICPhS, 1871-1874.
[16] Suraj Bhan Singh. Hindi bhasha: Sandharbh aur
Sanrachna. Sahitya Sahakar,1991.
[17] Whitney, W.D., 2002, History of Sanskrit Grammar,
Sanjay Prakashan, Delhi.
IJCSI
IJCSI CALL FOR PAPERS JANUARY 2010 ISSUE
The topics suggested by this issue can be discussed in term of concepts, surveys, state of the
art, research, standards, implementations, running experiments, applications, and industrial
case studies. Authors are invited to submit complete unpublished papers, which are not under
review in any other conference or journal in the following, but not limited to, topic areas.
See authors guide for manuscript preparation and submission guidelines.
Accepted papers will be published online and authors will be provided with printed
copies and indexed by Google Scholar, CiteSeerX, Directory for Open Access Journal
(DOAJ), Bielefeld Academic Search Engine (BASE), SCIRUS and more.
Deadline: 15th December 2009
Notification: 15th January 2010
Online Publication: 31st January 2010
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Evolutionary computation
Industrial systems
Evolutionary computation
Autonomic and autonomous systems
Bio-technologies
Knowledge data systems
Mobile and distance education
Intelligent techniques, logics, and
systems
Knowledge processing
Information technologies
Internet and web technologies
Digital information processing
Cognitive science and knowledge
agent-based systems
Mobility and multimedia systems
Systems performance
Networking and telecommunications
x
x
x
x
x
x
x
x
x
x
x
x
x
x
x
Software development and
deployment
Knowledge virtualization
Systems and networks on the chip
Context-aware systems
Networking technologies
Security in network, systems, and
applications
Knowledge for global defense
Information Systems [IS]
IPv6 Today - Technology and
deployment
Modeling
Optimization
Complexity
Natural Language Processing
Speech Synthesis
Data Mining
All submitted papers will be judged based on their quality by the technical committee and
reviewers. Papers that describe research and experimentation are encouraged.
All paper submissions will be handled electronically and detailed instructions on submission
procedure are available on IJCSI website (http://www.ijcsi.org).
For other information, please contact IJCSI Managing Editor, ([email protected])
Website: http://www.ijcsi.org
© IJCSI PUBLICATION 2009
www.IJCSI.org
IJCSI
The International Journal of Computer Science Issues (IJCSI) is a refereed journal for
scientific papers dealing with any area of computer science research. The purpose of
establishing the scientific journal is the assistance in development of science, fast
operative publication and storage of materials and results of scientific researches and
representation of the scientific conception of the society.
It also provides a venue for researchers, students and professionals to submit ongoing research and developments in these areas. Authors are encouraged to
contribute to the journal by submitting articles that illustrate new research results,
projects, surveying works and industrial experiences that describe significant advances
in field of computer science.
Indexing of IJCSI:
1. Google Scholar
2. Bielefeld Academic Search Engine (BASE)
3. CiteSeerX
4. SCIRUS
5. Docstoc
6. Scribd
7. Cornell’s University Library
8. SciRate
9. ScientificCommons
© IJCSI PUBLICATION
www.IJCSI.org

Full-Text - International Journal of Computer Science Issues

Transcription

Similar documents

Project A Registration for Parents 1. Begin at the WCATY homepage

pharmacy - Cooper

Franz Schubert Impromptus, Moments Musicaux, Klavierstücke for

Kinetico Yellow Page Program by How the Program Works in Five

KICKS AND GEAR RUNS THIS KICKS AND GEAR

Pimp my memory - Temple University

[email protected] [email protected]

Studs Terkel, with Tony Parker

bookweek edited.cdr - Navy Children School

ce1"f[,Caf~ w~re. ra.Y8I~ 1336.