A Mircroblog Recommendation System Based On User Clustering And Semantic Dictionary

Nowadays, microblog, as an information sharing and dissemination platform, has been widely used. How to help users retrieve posts they are interested in from massive amount of data becomes a very challenging issue. Some microblog recommend system attempted user-based recommendation. However, their effectiveness is limited by the short length of blog posts and vast diversity of user interests.In this paper, we implement the recommendation system based on micorblog, users can subscribe the post which they are interested in, system will recommend the related posts to users combined with user interest. On one hand, the recommendation system has specific target, so it has better recommendation result; on the other hand, the system can gather similar posts to form a topic which user is interested in.This paper’s recommendation system is based on user clustering and semantic dictionary. First, we establish user interest model through the collection of posts which user ever published and do the user clustering. Then, we collected the posts which is posted by users in the same cluster, do the similarity calculation with the psot which user subscribes, and recommend the post which has higher similarity to the user.Experimental results show that the improved K-Means clustering algorithm can achieve a better accuracy and running time. In the similarity calculation algorithm based on semantic, we add factors of weight and order of words, and achieve better discrimination and accuracy. Finally, we build a microblog system and implement recommendation function with the help of improved K-Means clustering algorithm and improved similarity calculation algorithm.