Research On The Techniques Of Discovery Of Opinion Leaders In Microblogger's Communities

In recent years, micro-blog is an emerging form of social network service, which leads to a huge revolution on the entire Internet. Micro-blog service narrows the distance between people. These users who uses micro-blog services can obtain a brand new method of the cooperation and communication. It is a novel way to generate the massive real-time and all aspects of the information that every person in the micro-blog network can play a role as the interface of the information communication. Therefore, the relationship between users and the tweets by retweeted constitute a large and complex network.In the process of the evolving the micro-blog network, it is a social network phenomenon that some small communities also come into being. Generally, the inside of the communities has the strong correlation and relation. If we can accurately divide into various communities,which is a further improvement of the quality of micro-blog service. With the comparison between substantial community detection algorithms, label propagation algorithms has proven to be an extremely fast method for large scale community detection and in wide range of the application. Random (node) access orders within the algorithm severely hamper its robustness,and consequently also the stability of the identified community structure. We propose a precedential label propagation that counteracts for the introduced randomness by increasing propagation preference. The results by experiments confirm that precedential label propagation is significantly more robust than label propagation.On the other hand, we have a research on the strategy of users' ranking. Most previous studies are based on a single factor including relationship between users or the amount of the tweet by users. However, an influential user has not only a number of followers, but also a mass of valuable information by his post. Our algorithm focus on the ranking strategies of users in terms of two major factors (posts and followers). Due to the character of the high real-time in micro-blog network, we treat the timeliness as a measure of an influence factor of a tweet.