Research On Evolution Mechanism Of Network Public Opinion Based On Topic

The openness and interactivity of Internet make the information spread more widely and faster. Once spreading through Internet, the influence of some hot issues, concerned events or sensitive topics will be multiplied, and even push the government, enterprises, organizations or individuals to the teeth of the storm in a moment. The monitoring and management of network public opinion has always been the subject of many scholars, some achievements have been obtained. This paper will explore the evolution rule of network public opinion from the perspective of topic, and will arrange and organize large amount of comprehensive information on Internet reasonably, trying to research the rule of online topic evolution and spread from a finer granularity, and offer certain theoretical foundation and guidance of researching methods for monitoring and management of network public opinion. The paper consists of following aspects:(1).The extraction of network public opinion topics. Based on the solid theoretical foundation of PLSA statistics, a method to extract network public opinion topic is put forward, and a large amount of network public opinion topics are extracted successfully through actual cases analyzing. Then the statistical features of the topic are used to accurately describe the semantic relationship between topics, texts and words, and the feature words and the contribution of the feature words are used to definite the topic quantitatively.(2).The evolution of network public opinion topics. Firstly, a statistical evolution model of network public opinion topics is constructed, by computing the degree of association between the feature words defining the topics, and using social network analytical tools to draw the network public opinion topics evolution path. The research finds that certain contacts exist between topics with even no literally association, and the contacts consistent with the co-occurrence probability of feature words; Secondly, a dynamic evolution model of network public opinion topic is constructed, and the system of network public opinion topics is treated as a dynamic network, through analyzing the microstructure of dynamic network and empirical researches, the dynamic evolution features of online public opinion topics is studied, including topic evolution path, hot topic evolution path and the concerned topic evolution path, and thus the evolution mechanism of online public opinion topic is revealed more profoundly.(3).The spread of network public opinion topic. Firstly, the spread mechanism of topic is analyzed according to the classical spread model SIR, and through simulation experiments, the influence of factors(user influence, user interest, interpersonal interaction, social memory effect) which affect the dissemination and diffusion of a topic is analyzed. Based on this, the conception of topic relevance is introduced, through the research of influence of topic relevance degree on factors relevant to the spread of topics based on multi-topics environment, a multi topic spread model is proposed. Some other experiments on topic correlation, multi topic occurring time interval and the influence of initial dissemination node on each topic and whole public opinion spread are conducted. Results of these simulation experiments indicate that, the topic spread model proposed in this paper can reflect the real topic evolution law and has certain practical significance.