The Establishment Of Microblog Hypernetwork Model And The Recognition Method For The Key Nodes

The coming of Internet age made the microblogging social network become the main carrier of the public opinion dissemination. The convenience and frequency of information exchange on virtual social network caused the attention of many scholars. Especially on the recognition of key nodes which have more impacts on the microblogging network structure and function in a larger extent, has become the research hotspot in the field of complex network.Due to the fact that homogenous network model in existing researches can't represent the real network condition and based on the analysis on research status and theoretical basis of the complex network, this paper combines the function and transmission characteristics of microblog and firstly introduced the hypernetwork model for microblog. It establishes the hypernetwork model from social, information, topic and spread wiliness these four dimensions.Traditional key node identification method is still based on the homogenous network model. Because of the multi-level attribute of hypernetwork, this paper puts forward the quantitative subnet node method based on microblog hypernetwork model, builds a kind of topic feature extraction method based on extensible LDA model and utilizes Chinese emotion ontology to identify the emotion subnet nodes to extend the current quantitative research based on network structure to hypernetwork research based on automatic modeling.For the problem about how to identify the key nodes under hypernetwork environment, this paper builds the hyperedge sorting algorithm(HyperEdgeRank) based on information dissemination effect, the topic semantic similarity and emotional consistency to calculate all hyperedge of user nodes and sort them to realize the recognition of key nodes.At the end, the paper uses the data from the real sina microblog and does experimental verification on topic extraction method and key node identification method. The experimental results show that topic extraction method based on the extensive LDA model can make up for the traditional LDA model in the insufficiency of the interpretability of the topic, the Chinese emotion ontology can extract the willingness node efficiently and the key node identification method can effectively and accurately identify key nodes in weibo hypernetwork, which also has the stronger application value.