Micro-Blog Friend-Recommendation Based on Topic Analysis and Circle Found

摘要

Because of the increasing popularity of Sina micro-blog, its data volume gets larger and larger. Friend recommendation gets harder. Users' behavior on Sina micro-blog reflects their value and interests. People who have similar interests are more likely to become friends. In view of the above-mentioned facts, we build micro-blog topic model based on users' operations and the concept of time slices. Then calculate the user similarity based on topic probability distribution that we get through the topic model. After that, clustering the users and getting social circles. Recalculating the user similarity based on circle structure and calculates user's trust degree of other users. In the end we can finish the friend recommendation based on user similarity and trust degree. Experimental results show this algorithm is better than traditional methods.