Design And Implementation Of PublicSentiment Detection System Based On Short Text Classification

The advent of network technology impacts on traditional media firstly. Social media, such as Facebook, microblog, Wechat show explosive growth, how to keep the influence of the traditional media and the monopoly in the press? Traditional media have to take the network data into the core resource of news production,then seek development and opportunities in the social transition period.This thesis studies the short text classification technology, and based on this technology,we will design and implement a public opinion information detection system. Firstly,this thesis describes the current situation as well as the development of text categorization, Liblinear, LibShortText classification tools, and the relationship between them.then introduces other technologies to build the system, including web crawler, and the Chinese word segmentation system called ICTCLAS2015.Secondly, the thesis make a requirement analysis of public opinion information detection system, including the system functional requirements, system performance requirements,.Based on the results of the requirement analysis, the thesis gives the detailed design of the subsequent network public opinion information detection system. Finaly, according to the detailed design, we have implemented the public opinion information detection system.The thesis completed the public opinion information detection system based on LibShortText.The system have four functions,including data collection, data storage, data analysis and the results showing. In order to verify the effectiveness of this system, we use the data of Lanzhou water peculiar smell on sina microblog to verify system, and achieve good effect. Currently, the system is in the testing phase.