TOPIC-VECTOR BASED USER MODEL FOR SOCIAL TAGGING SYSTEMS

According to the effect of enriching semantic information, social tagging systems have been regarded as novel information source for modeling user in personalized recommendation. Till now, most researchers construct the user model using weighted tag-vector. Although the simple and intuitively reasonable it is, the weighted tag-vector model has drawbacks including data sparsity problem and semantic ambiguity problem. In this paper, a topic-vector based user model is presented to solve the data sparsity problem and semantic ambiguity problem. With the discussion of the presented experiment, the validity of the modeling method was verified.