Study On Emotional Intensity Of Network Public Opinion Based On Semantic Web In Big Data Environment

With the popularity of the Internet,more and more Internet users pay attention to public opinion events through the network media.People browse public opinion news which they are interested in and express their views and feelings through the network.Due to the characteristics of openness and fictitiousness of the network,the spread of negative events speech may even affect the stability of the society.Because the large number of public opinion events are occurred in the same period,and every public opinion event also includes massive text information,the relevant departments need guide the network public opinion events timely and correctly.Therefore,it is essential to determine the emotional intensity of public opinion events.Based on the semantic syntactic analysis of public opinion text and text mining technology,this paper constructs the model of network public opinion emotion intensity.The specific research contents of this paper include the following aspects:1.The method of extracting text keywords is studied and improved.Aiming at the similar characteristic between data and the network public opinion,the improved core technology MapReduce model is used to mine the text of network public opinion.Map and Reduce function are combined to form the combiner.Then the combiner is used to calculate text feature weight and extract text feature keywords.So,the accuracy and efficiency have been improved.Then,based on MapReduce matrix-vector multiplication model,it can get hot public opinion events.2.Based on the sparse feature vectors of news headlines,the K-means clustering algorithm is improved.In the process of clustering analysis of news text,by understanding the randomness of initial clustering center for traditional K-means clustering algorithm,and the brief general characteristics of news headlines,the initial clustering centers are selected,which are based on the sparse feature vectors of news headlines.And then they are combined with K-means clustering algorithm,the clustering effect can be improved and validated by experiments.3.This paper builds the model of network public opinion emotional intensity from the objective and subjective aspects.First of all,the emotional dictionary which includes the emotional polarity and the emotional intensity is constructed according to the existing emotional dictionaries.The new dictionary includes the dictionaries of the modifier.And then calculate the emotional tendency of text based on the newdictionary.On this basis,it can build index of public opinion model from objective and subjective aspects.Through mining hot public opinion events,this paper analyzes the emotional strength of the network public opinion.This paper has a certain reference on the network of public opinion early warning research.