The Research On Key User Discovery Of Microblog Based On Signed Network

With the rapid development of Internet technology and the popularization of social networking,micro-blog users show an explosive growth and micro-blog information also shot up.In this case,find the key users timely and accurately in information communication,are of great significance in both advertising and marketing which need speed up the information dissemination,and rumor control and public opinion early warning which need suppress the information dissemination.But now micro-blog users finding areas still exist the following problems: 1)most of the existing micro-blog key users finding are faced unsigned network,ignoring the support or opposition relations between users,so that they are unable to obtain the accurate results;2)the existing key user finding methods are lack of ways to deal with the same problem in directed signed network,so that they can not meet the needs of key user finding to the directed signed network in micro-blog;3)the existing key users finding rules are similar to them in opinion leader recognition,which select the high ranking users as key users.They are unable to meet the various requirements in the practical applications.In view of the above problems,the core work of this paper mainly includes three parts:(1)A signed network model is constructed,based on micro-blog’s emotional polarity.Combined with the emotion dictionary and semantic rules could classified micro-blog forward text sentiment polarity.The positive and negative results are corresponding the "+" "-" edges of signed network.While the normalized weighted results of active users and user confidence are corresponding the node weights of signed network.These build a directed signed network on micro-blog.And the contrast experiment proves the superiority of this method used in emotional polarity classification.(2)The traditional PageRank algorithm is improved to make it used in the directed signed network.Using the node weight vector of signed network to modify the state transition probablity matrix to get the right positive / negative PR value.Combined with the influence of the degree of acceptance and unpopularity on the results,we get the final positive and negative results of the nodes,and the experiments shows the superiority of the improved effect.(3)The key user discovery rules are formulated.According to the different roles of different key users played in information dissemination,combined with the results of positive and negative influence of nodes,three key user discovery rules are designed,which are the positive communication affecting users,the negative communication affecting users and the controversial influence users.This paper constructs the signed network model based on micro-blog emotional polarity,and using the improved PageRank algorithm with node degree centrality to calculate the positive and negative influence of nodes in the network,finally identify three kinds of key users according to the key user discovery rules.The research realized key user discovery on signed network,which is a useful supplement to the identification of key user nodes in social network,as well as providing an important reference basis for decision-making in the fields such as new media marketing and public opinion dissemination.