Based On Web Food Public Opinion Analysis Key Technology Research

The rapid development of the Internet and all kinds of food intelligence has increased,more and more people through the web network to express their feelings and intelligence tofood view, from this web network will become food intelligence of public opinion distributioncenter. In this reality environment, for some hot topic and sensitive intelligence informationevent malicious incitement and deceptive report misled the broad masses of the people forfood intelligence correct understanding, in this kind of situation, it is necessary to the foodweb network intelligence topic and speech prison are effectively, help netizens correctrecognition of the food the immediate interests of the information and Internet users. Fromhuge web document data capture the public opinion in dynamic, it to stable society,constructing the harmonious and sustainable development of the society has the importantinstruction meaning. Therefore, based on the Web for food analysis key technology research ispublic opinion a pressing issue.This paper analyzes the information acquisition food public opinion technology,information technology and public opinion information analysis pretreatment technology for acomprehensive introduction and analysis, and focus on the food to the analysis of publicopinion two key techniques--text feature selection and characteristic parameters optimizationclassification problems:(1) The paper feature selection problems of research. The text characteristic selection ofmutual information considered rare low frequency Word of the mutual information than theuse of the condition of high, but MI did not take into account in the documentation of thenetwork in probability, make mutual information to choose and rare low-frequency words.Due to the high frequency noise words, missed the influence of factors, make its classificationeffect and feature selection is not ideal accuracy. In view of this limitation, this paper putsforward the average mutual information based on the characteristics of the improvement ofthe method. The experimental results show that the improved method can effectively solve theproblem the malpractices, ideal effect.(2) The paper to feature classification parameter optimization for research. The supportvector machine (SVM) model parameter selection Through the study of ant colony algorithmand the algorithm of support vector machine, and the combination of ACO-SVM algorithmare given. The experiment proved that ACO-SVM classification algorithm enhances the efficiency and accuracy of classification.