Design And Implementation Of Network Public Opinion Monitoring System Based On Clustering

With the development of science and technology,the Internet has gradually become the largest information exchange platform in China,which contains a lot of public opinion information.Monitoring the public opinion information and ensuring China's good network environment have gradually become one of the important contents of network management.In this context,this paper introduces the technology of data mining to realize the analysis of network public opinion,including three important functions,i.e.the theme analysis of text content,the clustering of hot topics and the early warning of topic heat.Through the research of this paper,a network public opinion monitoring system suitable for network supervision in China is realized.The main contents show at follows.First of all,we carry out a demand analysis on the network public opinion monitoring business process by combining with network management,the network public opinion monitoring management can divide into four kinds of functions,namely information collection,public opinion analysis,public opinion compilation and public opinion monitoring alarm.According to the role classification in the application of the system,we draw the corresponding business flow charts and functional use case diagrams respectively.Secondly,we carry out a system design according to the function demand analysis in this paper.In the function module of information collection,we introduce the network crawler technology to search information.In the public opinion analysis function module,we introduce K-Means Clustering analysis algorithm.It can transform the collected text information into empowering eigenvectors of keywords,and get the corresponding clustering center,that is,the vector representation of hot topics.It can monitor the public opinion by calculating and counting the heat and diffusion speed of topics.In addition,we also carry out the conceptual design and logical structure design of the back-office database of the system,and then draw the corresponding E-R diagram and the specific database form design results.Finally,we implement the system by using J2 EE technology architecture,and give the core code of different function modules.We test the 13 functional use cases in the system respectively,and test the performance of the system by using LoadRunner and other software,it proves the effectiveness of the system.The system developed in this paper monitors the high frequency themes appearing in the network,and automatically monitors the frequency of keywords and the changes of heat,which greatly reduces the working intensity of the staff.