Automated Collaborative Filtering in the Presence of Rating Imperfections
Thanuka Wickramarathne, Kamal Premaratne, Miroslav Kubat and Dushyantha Jayaweera
Problem
An example of a typical application is an e-commerce where customers rate items and receive automated recommendations based on detected similarity patterns. One of the major problems encountered by conventional automated collaborative filtering (ACF) algorithms is data imperfection (limited statistics, subjective judgment, and so on). Conventional algorithms either completely ignore imperfect user ratings or utilize some imputation mechanism to remove the imperfections (e.g., fill-in the missing entries). Neither strategy produces acceptable result, especially when a large fraction of the data is imperfect and/or little information is available about the reason for and the mechanism driving the imperfections. This is one main reason for existing ACF algorithms not being widely utilized in applications where data imperfections are commonplace, such as medical/healthcare data, homeland security and defense applications, etc.
Solution
The present algorithm is specifically designed to work in the presence of imperfect data. It is a Collaborative Filtering method based on Dempster-Shafer belief theoretic framework that can represent a wide variety of data imperfections, propagate them throughout the decision-making process without the need to make simplifying assumptions.
Competitive Advantage
One of the major problems encountered by conventional automated collaborative filtering algorithms is imperfection of available data. The method of the present invention was specifically developed to work in the presence of imperfect data.
Applications
The invention provides a method of data pattern analysis in support of decision making
Patent Status
International patent application entitled AUTOMATED COLLABORATIVE FILTERING IN THE PRESENCE OF RATING IMPERFECTIONS was filed on July 16, 2009.
Licensing Opportunity
We are looking for a commercialization partner with capabilities in product development, sales, and marketing. An exclusive worldwide license is available.
About the Inventors
Thanuka L. Wickramarathne is pursuing his doctoral studies in the Department of Electrical and Computer Engineering at the University of Miami. His interests include uncertainty modeling and knowledge discovery from imperfect data. He received his B.Sc. in Electronics and Telecommunication Engineering from University of Moratuwa, Sri Lanka, in 2006.
Kamal Premaratne is a Professor in the Department of Electrical and Computer Engineering at the University of Miami. His research interests include evidence fusion and resource management in distributed decision and sensor networks, DS belief theory, knowledge discovery from imperfect data, and network congestion control. For his research work, he has received the Mather Premium and Heaviside Premium of the Institution of Electrical Engineers (IEE), London, UK, and the Eliahu I. Jury Excellence in Research Award of the College of Engineering, University of Miami. He has served as an Associate Editor of the IEEE Transactions on Signal Processing and the Journal of the Franklin Institute. He is a Fellow of IET (formerly IEE) and a Senior Member of IEEE. Dr. Premaratne received the B.Sc. degree in Electronics and Telecommunication Engineering (with First-Class Honors) from University of Moratuwa, Sri Lanka, in 1982, and his M.S. and Ph.D. degrees, both in Electrical and Computer Engineering, from the University of Miami, in 1984 and 1988 respectively.
Miroslav Kubat is an Associate Professor in the Department of Electrical and Computer Engineering at the University of Miami. In the fields of machine learning, neural networks, and data mining, he has published more than 80 refereed papers. He also co-edited the books Machine Learning and Data Mining: Methods and Applications (Wiley, 1998) and Machines that Learn to Play Games (NOVA Science Publishers, 2001). He is on the editorial boards of the three journals Intelligent Data Analysis; Informatica: An International Journal on Computing and Informatics; and Applied Artificial Intelligence.
Dushyantha T. Jayaweera is a Professor in Clinical Medicine in the University of Miamis Miller School of Medicine, where he has a Clinical Trials Unit. He is in charge of the HCV/HIV co-infection clinic; he is also the Physician-in-Charge for HIV in HIV/Liver Transplant Program. He is the Chairman of the University of Miami Medical Ethics Committee. His research interests include HIV new drug development, drug abuse and hepatitis-C. Dr. Jayaweera obtained his M.D., and his M.S. in OBGYN, from the University of Colombo, Sri Lanka, in 1977 and 1981, respectively. He completed the MRCOG (UK) in 1984, and his residency program at the Loyola and Hines VA program in Internal Medicine. He was awarded FRCP by the American Board of Internal Medicine in 2005.
Selected References
Baum, MK, Jayaweera, DT, Duan, LR, et al (2008). Quality of life, symptomatology and healthcare utilization in HIV/HCV co-infected drug users in Miami. J of Addictive Diseases, 27(2).
Boyle, A, Jayaweera, DT, Witt, MD, et al (2008). Simplification to a once-daily Efavirenz-based highly active antiretroviral therapy regimen from a twice-daily or more frequent regimen improves adherence while maintaining viral suppression. HIV Clinical Trials, accepted.
Waldrop-Valverde, D, Jones, DL, Jayaweera, et al (2008). Gender differences in medication management capacity in HIV infection: The role of health literacy and numeracy. AIDS and Behavior.
Jayaweera, D, Espinoza, L, Castro J (2008). Etravirine: the renaissance of non-nucleoside reverse transcriptase inhibitors. Pharmacother, 9(17):1-12.
Subasingha SP, Zhang J, Premaratne K, Shyu M-L, Kubat M, Hewawasam KKRGK (2008). Using association rules for classification from databases having class label ambiguities: A belief theoretic method. Chapter in Data Mining: Foundations and Practice, Eds: Lin TY, Xie Y, Wasilewska A, Liau C-J, Studies in Computational Intelligence (Springer-Verlag), 118.
Wickramaratna K, Kubat M, Premaratne K (2008). Predicting missing items in shopping carts. IEEE Trans on Knowledge and Data Eng, accepted.
Wickramarathne, TL, Premaratne, K, Kubat, M, Jayaweera, DT (2008). CoFiDS: A belief theoretic approach for automated collaborative filtering. IEEE Trans on Knowledge and Data Eng, in review.
