Microblog Sentiment Analysis Model Based Hot Topic Detection On Public Opinion

With today’s rapid updating of information technology and development, Internet has become released, access and transmission of the main carriers of information. Internet public opinion, as the popular will and the description of the effect on its rulers and the political orientation laws, is the comprehensive reflection of the popular will. With network as the carrier and event as the core, the internet public opinion is the synthesis of the expression, communication and interaction of most internet users’emotions, attitudes, opinions and ideas, as well as the subsequent influences. On the other hand, the internet public opinion, as an important part of the society public opinion, is the mapping of the society public opinion in the internet space. For the relevant government departments, discovering the internet public opinion information accurately and operating supervision and guidance on it, will greatly contribute to the social harmony and stability in practice.Aiming at the characteristic of microblog data, this paper presents a model based on microblog sentiment analysis to find out internet hot topics. And through this way we can effectively monitor abnormal or unexpected events, which has positive implications for internet public opinion research. Firstly, we dealt with the data of a given topic over time, classified them with classifier based on smilies and emotion words, and obtained three types of microblog-positive emotions, negative emotions and extreme negative emotions. Combined with psychological knowledge, we analysed data graphs of microblog later. Through the proposed event extraction method based on word co-occurrence graph, we extracted keywords of the extreme negative emotion, and used word co-occurrence diagram to identify network hot spots within this time.Experimental results show that for the microblog containing emotional knowledge in a topic in a certain period, using the model of microblog sentiment analysis based on this paper can accurately find out the network hot issues of this time.