A Semantic Category Recommendation System Exploiting LDA Clustering Algorithm and Social Folksonomy

According to a widespread use of the internet, the amount of data generated from various Social Network Services (SNSs) is increasing day by day. Thus, it has become necessary to categorize data for users to efficiently access to the desired information. However, most of the web sites do not provide the categorizing service. Even a few sites that offer the categorizing service do not support user-oriented automatic category generation with a high-quality performance. In addition, there are several limitations in an analysis of large amounts of data because of a high dimension of vectors when clustering data sets for the category generation. This paper proposes a system that provides users with a service for recommending categories by utilizing social folksonomy with clustered data. Further, a method to reduce the dimension of vectors by removing meaningless words in the contents is introduced.