Research On Microblog Recommendation Algorithm Based On Topic Heat

The rapid spread of the Internet allows users have access to massive information. But it becomes difficult for them to search for information effectively and results in the problem of "information overload". Recommendation system is one of the most effective tools to solve such problem. In recent years, scholars have made some improvements in the research and application of the recommendation system. There are still a series of problems about the recommendation algorithm which is applied to the social network, such as the inaccurate calculation of heat value, cold start problem, etc.Due to the inaccurate calculation of heat value, this paper presents a kind of calculation method which is based on specific user’s role. Due to the cold start problem in recommendation algorithm, this paper presents the micro-blog topic recommendation algorithm combining artificial bee colony. And the micro-blog topic recommendation prototype system is designed and implemented by applying these calculation methods. The work in this thesis mainly includes the following three aspects.(1) Researched the hot topic discovery method of public opinion analysis, and proposed a method to calculate heat value which is based on specific user’s role.By taking publisher,attention,time,numbers of forwarding and numbers of comment into consideration, this method can help us calculate the heat value of the topic through setting specific user’s weight. The experimental results show that this method not only can detect the hot topics effectively, but also can help regulatory authority monitor and manage the micro-blog hot issues.(2) Researched the artificial bee colony algorithm and collaborative filtering recommendation, then proposed the micro-blog topic recommendation algorithm combining artificial bee colony.Firstly, this algorithm constructs the fitness function based on the four factors: heat of a topic, user characteristics, user preference, starting time. Then, we solve the fitness value by the constructed fitness function. Finally, micro-blog topics are recommended to the users according to the fitness value. The experimental results show that this method can solve the cold-start problem that exists in the collaborative filtering recommendation algorithm efficiently and improve the precision of the recommendation.(3) Designed and implemented the topic recommendation of micro-blog prototype system based on hot topic valueWe apply the improved recommendation method to the system. This system includes five modules which are data collection, data processing, topic detection, topic discovery and topic recommendation. We realized method of hot calculation based on particular user roles and the algorithm of topic recommendation combining artificial bee colony. The results show that these two methods can effectively improve the accuracy of the micro-blog topic recommendation.