Towards Efficient and Scalable Memory-based Collaborative Filtering

Speaker: TAM Ming Wai

Abstract

It is common for one looking for interesting information, but the task is not easy as there is great amount of information. Some domains, such as movies, rely on professional reviewers. This solution is clearly not good enough, since the taste and interest are different among people. Memory-based Collaborative Filtering is popular in personalized preference prediction, which is aimed at the above application. However, the efficiency of Memory-based CF is low, which is a problem in large database. Alternatives to Memory-based CF suffer from other overhead. This talk will give an overview on CF, present a solution given in a SIGIR2002 paper, and outline some possible further improvement to the CF algorithm.