Research On Rapid Topics Discovery And Influence Accessment Method In Microblog With Public Emotional Engagement

With the rapid development of mobile social network, the emergence of a large number of short text messages keep growth in a geometric way represented by micro-blog. With unstructured, massive, real-time and media characteristics, micro-blog gathered a lot of valuable information relates to people’s daily life, it’s known as the topic. Because micro-blog text is different from the traditional news, unshackled with the title, keywords and other fixed format, it’s also bear the functions of message, blog, instant messaging and others in the era of mobile Internet. The influx of large number short text messages at the same time, bring new challenges to the topic information capture and refining. As a kind of mobile social networking tools, the demand of novelty, personalized in topic information is more and more important in public attention, recommended topics to user at the right time and place has gradually become an important means to maintain active degree of users. Traditional topic propagation methods no longer meet the personalized requirements of users. Therefore, for a better services to users, how to quickly and extract information, discovery efficiently and spread micro-blog information timely and accurately has become an new research hot spot.This paper presents a new method of micro-blog topic discovery quickly and influence assessment with public emotional engagement. The main research content includes the following several aspects:First, in view of the vast amounts of micro blog, this paper presents a method of micro-blog topic discovery quickly based on public emotional engagement. The method mainly constructed by the building of emotional lexicon, emotion text extraction and micro-blog topic discovery. First part, based on the large-scale micro blog corpus and the three famous emotional word sets, using TFDF values and double word index Hash table to achieve specific emotional word library building; Second part, we use Sigmoid function to mine emotional dense phase and extract emotions text. Third part is an iterated and optimized process, based on improved Fuzzy clustering algorithm, the nominal entity table was established after the text reduction. For better efficiency and quality, public emotion represent by emotional word was used in topic discovery process to reduce a large-scale micro-blog text. Second, according to the interdependent relationship between emotional words and text number, this paper put forward the concept of emotional intensity period. The mining algorithm of emotional intensive period considering the propagation characteristics of micro-blog, micro-blog retreat, comments and other influence factors. The using of Sigmoid function could effectively suppress the high frequency commercial micro-blog impact on the numerical results, and enhance the quality of topic discovery in mining intensive period of public emotional investment,Third, with the rapid development of mobile internet, more and more people cannot live without social network. To solve the problem of user node in the differentiation of social network effect, this paper presents a kind of user node influence measurement method based on public emotional investment. To research user influence assessment method in new network structure, emotional words was used to define the topic emotional concentration values as the basis. Then, topic emotional concentration values could seen as user interest value to avoid impact of topic interest differences in each user. Proved by experiments, we obtain a better precision influence evaluation.