Research On Topic Recognition And Sentiment Analysis Technology For Sensitive Information On The Internet

With the rapid development of Internet and its equipment,the way people access Internet information is shifting from passive acceptance to active creation and sharing.The openness and convenience of the network make more and more network users choose to publish some information on the Internet,express their views,or just vent emotion on the Internet.This information plays an important role in the formation and dissemination of public opinion,but there are also potential security threats.Therefore,how to mine potential topics in a large number of Web Texts and analyze the emotional tendency of published texts is a very valuable research topic.This paper will do the following research for web text:1)The current topic recognition model has disadvantages,which is that the recognition rate of some vocabulary or domain vocabulary with sensitive tendencies is not high,and the generated keywords are not accurate.In order to solve the above problems,this paper proposes a network text information subject recognition model based on Key word weighted-LDA model.On the one hand,this method embeds the keyword vocabulary into LDA topic model to improve the semantic understanding and recognition ability for domain words,and improve the quality of generating topic words.On the other hand,it can also improve the relevance between topic words and related topics and find more fine-grained topic words.The experimental results show that the Key Word weighted-LDA model can effectively improve the quantity and quality of the recognition of domain keywords and enhance the recognition ability of topic recognition.2)Research on sentiment analysis of network text.In this paper,the task of sentiment analysis is carried out under specific topics.Based on this,this paper proposes a text sentiment analysis method which integrates subject semantics information.This method integrates topic semantics information into text representation by constructing a neural network.In order to increase the contribution of emotional words to text emotion,the Attention mechanism is introduced into the neural network,the context-aware vectors are introduced to calculate the weighted weights of each word.The experimental results show that the proposed model can effectively improve the accuracy of emotional analysis results.In this paper,the topic recognition and sentiment analysis technology of Web text is studied.It is beneficial for the relevant public opinion monitoring departments to better grasp the network public opinion,understand the social public opinion,and promptly respond to the demands of netizens.It is also conducive to strengthening the benign interaction between the government,the media,and the netizens,and building a harmonious and orderly network environment.In addition,this paper also has a certain promotion effect on the development of related technologies.