Font Size: a A A

Research And Implementation Of System

Posted on:2015-03-22Degree:MasterType:Thesis
Country:ChinaCandidate:G T OuFull Text:PDF
GTID:2308330467475516Subject:Software engineering
Abstract/Summary:PDF Full Text Request
Network public opinion refers to the different views on the social problem of popular on the Internet, is a form of social public opinion, is through the Internet, held by the public, focus on some hot issues in real life there is a strong influence, tendentious statements and views. Strengthen the network public opinion information work is helpful to guide public opinion correctly, so as to continuously maintain social stability, harmony, is the primary network supervision need to focus on the work in the complicated environment in the new period, from this point on Web networks is proposed and the characteristic of the system based on content.This thesis based on analyzing the network public opinion of processing information acquisition technology, web crawler technology, data mining technology and other key technologies, grassroots oriented network monitoring real work demand, the design and implementation of network public opinion monitoring system, the system uses the DOM tree and the improved PageRank model processing page, lay a good foundation for text data treatment of late; in public opinion found on classification and cluster analysis based on the technology, this paper proposes a classification algorithm based on two layer structure, the calculation of feature selection and weighting algorithm is improved. After the experiment and the algorithm of the algorithm characteristics can improve the efficiency of the algorithm. In addition, the improvement of the traditional clustering algorithm, proposed the divisive hierarchical clustering algorithm of distributed data based on the division, and in the HADOOP platform to achieve data partitioning method based on maximum frequent term sets, through the simulation experiment can reduce the communication overhead. Finally, the development trends of comprehensive information and emotion index, development trend of a network public opinion over time, experimental results obtained from the experimental units of the system, prove the validity of system research and design.Application test shows that the system can better realize the network public opinion management requirements based on Web, realize the public opinion monitoring and perfect auxiliary decision makers to make accurate decisions, at the same time, the information of network public opinion supervision effective technology will decrease the work intensity, grassroots network monitoring in a large extent, improve work efficiency.
Keywords/Search Tags:Network public opinion, improved PageRank model, web crawler, divisive hierarchical clustering algorithm
PDF Full Text Request
Related items