Invited Speaker---Dr. Soojung Lee, Professor
Department of Computer Education, Gyeongin National University of Education, South Korea
Soojung Lee received her M.S and Ph.D. in department of computer science at Texas A&M University in 1990 and 1994, respectively. Since then she was working as a senior technical manager in Telecom Network Systems Division of Samsung Electronics Co. until 1998. There she was involved in developing telephone switching systems using X.25 communication protocol, network management systems for SDH equipment using CMIP based on TMN concepts, and management systems for ATM network of telephone exchanges, LAN, PDH, ATM, Frame Relay using SNMP. In 1998, she joined the faculty as a professor of Department of Computer Education in Gyeongin National University of Education. Her research interests include recommender systems, fuzzy logic and data mining techniques.
User-based Collaborative Filtering Using Fuzzy Clustering
Collaborative filtering is a well-known technique successfully used in various recommender systems. However, it suffers from major drawbacks of the scalability and the data sparsity problems, when the system makes a recommendation based on the ratings records of similar users. This study aims at solving these problems by exploiting user interest in movie genres to build clusters of users. We make a slight variation of Fuzzy C-means algorithm such that several centres, one for each genre, per cluster are maintained. Experimental results showed that the proposed strategy demonstrated performance comparable to a conventional method without using clustering and significantly better than a well-known clustering algorithm of K-means.