Term Co-occurrence Analysis And Opinion Leader Recognition Of Micro-blog

With the rapid development of the web 2.0, the concept of openness and sharing is deeply rooted in the hearts of the people. Internet users on the network platform for the use of social network has been greatly improved. The rapid development of information propagation speed in large data network, the public opinion on the micro-blog platform is becoming more and more social trend oriented. How to quickly and accurately find the topic of public opinion in a large number of short text messages becomes the primary objective of the supervision of public opinion, and the opinion leaders can effectively guide the direction of public opinion. The main work of this paper is focus on the identification of hot topics and opinion leaders, the specific work is as follows:(1) Traditional word co-occurrence detection methods in micro-blog news encounter the problems of high computational complexity, high time consuming, and low recall rate. An improved algorithm of word co-occurrence detection based on Rough Set is proposed in this paper aiming at solving these problems. The idea of a subject is expressed by the co-occurrence of words, and it builds a word co-occurrence matrix through word co-occurrence relation, and the maximum complete subgraph method for rigorous topic clusters. Finally, it extended the topic clusters based on Rough Set Theory. Experimental results carried out that the proposed method can effectively reduce the time complexity of the traditional algorithm, and accurately detect the core vocabulary of hot topics.(2) It can better reflect the hot spot of public opinion through the emotion information mining, and the negative emotion is easy to cause the serious social influence. At the same time emotional information can filter more irrelevant information and improve the quality of topic detection. An improved algorithm of word co-occurrence detection based on emotions is proposed in this paper. Based on the co-occurrence of the emotional elements, we establish the emotional co-occurrence matrix, and then we form a special emotional subspace model by clustering method. We divide the topic information by using the emotion subspace, and then extract the core words from the micro-blog information in different categories. The method can filter out most of the hot topic irrelevant information, and reflect the enthusiasm of the public opinion information. It is conducive to the monitoring of public opinion information, and improve the accuracy of the traditional algorithm.(3) Opinion leaders in the information dissemination are always to promote the development of public opinion, and the speed of information dissemination of opinion leaders faster than ordinary users, also cause greater influence. Detection of opinion leaders is of vital importance to the public opinion monitoring work, so the method for the detection of opinion leaders in a dissemination network of the specific topic is proposed. The information propagation network is established by using the forwarding relation. In the algorithm, the model combines influence evaluation method of PageRank on specific topics under the opinion leaders identification, and the user's influence is divided into its own authority and forward the user's support. Experimental results show that the proposed method can effectively detect the opinion leaders in a special topic.