Information Filtering and Personalization Services

Abstract

Personalization service is becoming one of the core services in digital libraries, and an exciting and challenge research area. In this chapter, we analyze several key technologies and the related works in information filtering and personalized services, and then present a content-based personalized searching algorithm and a probabilistic model to represent user interests, which is more effective than the vector space model by the experiments. To solve the data sparsity and scalability problems in collaborative filtering, we present new methods for similarity computation and instance selection. The experiments show it is higher predicted precision and performance than the others. Based on the above research results, we design and develop a prototype, TH-PASS, which provides personalized searching and recommending services.