Research On The Construction Of Index Systerm Of Emergency Network Public Opinion Monitoring

Since the 1990 s, the internet has won huge of popularity, which plays the role of the fourth media to voice netizens’ s voice in our country. With the increasing number of netizens, their opinions and behaviors always mean a serious threat to our society. Because of the rapid growth of the internet which has become an important way for the public to express opinions, the influence of network popular opinions is coming to be more and more huge.As a special form of network public opinion, the network public opinion of emergencies always burst with the accumulation of various social conflicts and hotspot issues. Along with the breeding and mutation of public opinion, the number of the related public opinion information which result from positive participation of network media and netizens, is becoming more and more, and even increasing explosively. At the same time, structural diversity and cultural differences of netizens, together with diversity of network media, makes network public opinion of emergencies becoming more complex and unpredictable. Timely monitoring and accurate analysis of network public opinion of Emergencies has become an important prerequisite for governance emergencies and correct guidance of public opinion.Building index monitoring system for public opinion of emergencies can be divided into the following steps: 1. Based on the study of the relevant national authorities and papers on public opinion monitoring, together with according to the preset principle, primaries get to be screened the original set of indicators of public opinion. The index set that can be divided into three levels covers sources, dissemination and audiences of public opinion. Besides the above mentioned, the last stage has a strong observability, objectivity and broad representation; 2. Using correlation analysis and principal component analysis combining analytical methods is logical and rigorous. Firstly, we exclude the correlated indexes significantly associated with evaluation to ensure that the screened index system obtains a lower degree of redundancy without sacrificing the index system’s representation by correlation analysis, and then obtain a few highly representative indexes screening from the original index set by principal component analysis using the linear transformation and the preset cumulative variance contribution rate threshold, which further guarantee lower redundancy, strong representation and lower complexity; 3. We fix the index weight of the screened index set using AHP of which Delphi and correlation analysis is combined creatively in the judgment matrix constructing. It both ensures the actual importance reflection and objectivity of index weight determination. 4. Model validation: Introducing another public opinion emergency, validation of rationality and feasibility of the screened index system can be taken by creative calculation of index correlation between Emergency Network Public Opinion Index and Baidu Opinion Index.