Research On Hot Topic Detection And Event Tracking In Network Public Opinion

With the rapid development of Internet, more and more people would like to express their thoughts, feelings and attitudes through network, the negative network public opinion will bring great threat to social public safety, and analyzing public opinion is of much significance to steadying social order and promoting nation development. So, the Hot Topic Detection and Event Tracking Technology is explored and a HaiSi Public Opinion Analyzing and Monitoring Management System is constructed in this paper.Due to its essential role in network security and social public security, research on public opinion analyzing is of great importance. A primary study of the key technologies involved in public opinion analyzing system is presented, including information acquisition, information preprocessing, hotspot detection, event tracking and public opinion prewarning:First, the information acquisition and information preprocessing technologies deal with the study of grabing News or BBS web pages at appointed scope by network crawler, also study extracting text and Chinese language segmentation to generate weighted vectors. The research of network crawler limitations and relevant improvement policies is emphasized in chapter 3 of this paper.Second, the hot topic detection and event tracking technology is the core problem of public opinion analyzing. The improved incremental clustering algorithm single-pass and auto-learning-based incremental classing algorithm 1nn are studied separately to implement public opinion new event, hot event, focuse event and sensitive event detection and tracking in chapter 4 of this paper.Third, the prewarning technology is the study of early warning which event will become a hot hopic before it real is by trendency analyzing public opinion influending factors. Here we make a study of applying ARMA-based forecasting model combining with multiple linear regression model in public opinion early warning system in chapter 4 of this paper. At last, a prototype system-Network Public Opinion Analyzing and Monitoring Management system is designed and implemented grounded on the research of Public Opinion Analyzing technology. Firstly, the designing of the main body frame is presented, and then the designing of each submodule are introduced separately. The architecture of the system runs on networking hosts with Unix/Linux operating system. It is competent for the task of public opinion information collection, analyzing, monitoring and early warning, and by which effectively assisting the artificial public opinion monitoring and reducing man-made interference.