Design And Implementation Of New Topic Detecton And Tracking Of Microblog Based On Emergency

Compared with traditional media, microblog with its short, quick and easy characters quickly has become a public message medium, and has grown into one of the main platforms for people to obtain and publish information. At present, microblog has developed a large number of user groups, where the great amount of information is published every day. The researches on topic detection and tracking are also transmitting from the traditional news to the microblogging platform. And emergencies happen about national security and social stability, which makes that topic detection and tracking is an important significance based on all kinds of emergencies in dealing with the crisis management. This paper focuses on microblogging platform and uses the topic detection technology to mine related topics about the emergency information, and spreads way of modeling research topic based on the microblogging network.This paper uses the research method by combining theory with practices, including domestic and foreign research situation, microblogging data collection, topic detection and emergency microblogging transmission network modeling. First of all, based on domestic and foreign research in data mining, clustering method and the complex network, the paper put forward the new points in topic detection and tracking for emergency by paying close attention to the developing trend of related research at home and abroad. Then, it collects the data sets from Sina microblog by designing the crawler for microblogging platform and extracting and preprocessing the microblogging data. On the basis of the complete data collection, it begins detecting the topics about emergencies. Because the content is brief and colloquial which lead to common text clustering method not being applied to the microblogging data effectively. Therefore, the paper choose the LDA algorithm as the topic detection method, and put forward the improved TC-LDA algorithm due to the background noise problem on the data set. In the end, this paper spread the way of the microblog communication network research by using the method of combining Petri net and the Agent, which can make it from both micro individual and macro system observing the transmission features of microblogging network.In one word, through the research on topic detection and tracking based on the microblogging platform, this paper design and implement a system including collecting microblog data, topic detection and modeling the microblogging network so that it can provide reference for the detection and management of emergencies.