Research And Implementation Of Analysis And Management Technology Of Network Public Opinion

With the development of Internet technologies, network has become an important place where public opinions are produced and expanded. Network public opinion can effectively present current concerned hot topics, thereby reflecting current social, political and economic development situation. Therefore, it makes great practical significance of grasping current dynamics and guiding the development of network public opinion by collecting and analyzing public opinion. This article is to make a study of the analysis technology of network public opinion, and based on which, the analysis and management system of network public opinion is designed and implemented. The following aspects are included in this paper:1、Collection and preprocessing of network information, which are the basis of research on network public opinion, are defined as seizing and processing a web page. The second chapter focuses on the basic idea of information collection and preprocessing technologies.2、Topic detection and topic tracking, key of network public opinion analysis, refers to topic clustering on unknown topics and topic classification on known topics. An improved clustering algorithm based on Single-Pass is proposed in the chapter3. The improved clustering algorithm has two major innovations:The first innovation is assigning higher weights to double proper nouns when extracting feature vectors, because of the better distinguish degree of proper nouns or double proper nouns. Secondly, that’s, during the similarity comparison, comparing the document title feature vector of all documents to get the division of parent class topics, and then comparing the document text feature vector of documents under the parent class topics to get the division of subclass topics, so as to hierarchically present the clustering results.3、Topic evaluation and early warning of network public opinion are ultimate goal of public opinion analysis. The evaluation method and evaluation process of network public opinion will be studied in chapter4. First, based on the existing network public feelings evaluation index system and statistic index value of hotspots in the past, proceeding from five dimensions as quantity, the time, the attention, the viewpoint and the concentration degree, a network hotspot evaluation index system of public feeling is constructed; And then, using the Analytic Hierarchy Process to determine the index weights of topic which is to be evaluated; After the quantization process, all index value can be calculated and the evaluation value will be obtained; Finally, the corresponding topics will be handled according the evaluation value and rating table of early warning.Finally, the analysis and management system of network public opinion is designed and implemented. After designing the overall framework of the system, we separately introduced the function and implementation of each module. Modules of collection, preprocessing, analysis, early warning and management are designed, and modules of collection, preprocessing and analysis are implemented in the system, which has practical significance to the network public opinion application.