Research And Implementation Of Emergency-objected Monitoring Algorithm Of Internet Public Sentiment

With the rapid development of WWW application,information dissemination using Internet as a newest kind of dissemination of information has deepen into the human's daily life,which is never substituted by the traditional media.Because the virtually and concealment of network causes consensus on the network more and more serious,how to guide the development of network public opinion and maintain social stability has become an important issue in current research.According to the analysis of the existing network public opinion research results,a common structure model of network public opinion monitoring system is provided in this thesis.Then each component in the model is introduced.Furthermore,the technologies applied in analyzing module about public opinion are summarized.The traditional classification algorithms like naive Bayesian are affected by the quantity of irrelevant features.In order to reduce the number of irrelevant features and improve the classification performance,a new algorithm is provided.The new algorithm used association rule mining algorithm,and obtain the relevance between words.Then it extracted the words those have little relevance between words.Using the remained words to compute,the precise of NB classification algorithm get improved.The experimental results also show the effectiveness of the method.Analyzing some existing emergency monitoring algorithms,it is easy to make a conclusion that those public opinions with a same central topic have high similarity.Therefore,a novel kind of emergency monitoring algorithm based on condensed chain is provided.Using TFIDF to calculate the weight of features,a new method of similarity measure between two different documents is defined.On this basis,this algorithm computes the density of each document in its' neighbors,then searches the central document set along with the condensed chain.Thus,the current public opinion focus topic set with high density make up into public opinion hotspot.Using the news from SINA Website,the experiment is made to obtain the hot News.With artificial analysis,these hot News are verified,which shows the effective of this new algorithm.Based on the existing public opinion monitoring system,this thesis gives a simple public opinion monitoring prototype system.Each part of the system is introduced in detail not only about the function but also its' key code.