Research Of Potential Opinion Leaders Based On The User Characteristics Under Social Network Environment

With the continuous development of Internet technology, especially in recent years, the social network sites grow popular, such as microblog, people prefer to communicate with others or express views and opinions on events through online social networking sites more and more. Online social networking sites have into people’s daily life. Sina microblog, for example, events recently about the livelihood of the people, corruption or others once be put forward, they will receive a large number of microblog users’ attention, and then the topic quickly heats up rapidly. If we do not make these events which happen online into supervise and control, and let them which are more dominant and authoritative just go on without limitation, it will be bad for our security of the entire network environment, and it has a bad influence to the realistic society indeed.All researchers of each field have made encouraging progress about mining opinion leaders so far, and summarize the results of these studies we can find that the methods of mining opinion leaders are mainly based on statistical methods, social network analysis method, and clustering analysis, etc. These research methods have very good effect on mining opinion leaders, but they are all using the thought pattern of “from back to before”, the study do not start until the event has already become hot, and this way is not consistent with the normal process of messages spreading on social network. This paper wants to find a new breakthrough about mining opinion leaders by studying, so we puts forward the concept of “potential opinion leaders”. “Potential opinion leaders” are the people or organization that exist in social network, and play an important role in the development process before ordinary events becoming hot. The research of “potential opinion leaders” takes advantage of “from front to back” thought pattern,which is more consistent with the normal way of thinking, and compared with the traditional opinion leaders mining pattern,it is more reasonable and pragmatic.The main contents of this paper include:1. In view of the methods of mining opinion leaders do not conform to the normal development of events online, this paper puts forward the concept of “potential opinion leaders”, it is according to the normal order of the development of events, and the information feedback is more timely and effective;2. In this paper, we demonstrate the rationality and practicability of the "potential opinion leaders", and expounds the reason why the core idea of PageRank algorithm and degree centrality can be applied to the research of "potential opinion leaders";3. In this paper, we explain the user characteristics, and put forward the research model of "potential opinion leaders" which based on the analysis of user characteristics, and then the rationality of the model is verified by experiment, it offers a new way for the network early warning.