A Hybrid Content-Based Filtering Approach: Recommending Microbloggers for Web-Based Communities

Content in microblogging systems is produced by hundreds millions of people and is definitely becoming a huge internet resource repository now. Previous research on social media recommender system often focuses on followee prediction by collaborative and content-based filtering. In this paper, we concentrate on recommending microbloggers for web-based communities with potential topics, by using a hybrid content-based filtering approach, based on top model clustering and TF-IDF cosine similarity. We evaluate our study using a dataset of SINA WEIBO, the largest micro blogging service in China, as well as sample online communities from RESEARCH ONLINE. Intensive evaluations are conducted so that the proposed hybrid filtering approach is effective and reliable.