Research On Hot Topic Detection In Network Public Opinion

Public opinion refers to the public’s social and political attitudes towards certain hot events in society.It is the sum of beliefs,attitudes,opinions,and emotions expressed by the masses about various social phenomena and problems.With the development of science and the extensive application of the Internet,the network has become an important platform for people to obtain information and express their opinions,and network public opinion has also become an important form of social public opinion.If the public opinion on the network is improper handled,it may result in a huge impact on society.Consequently,more and more scholars have been devoted to the study of public opinion.Research on hot topic of public opinion is an important direction of public opinion analysis,and it includes the topic detection of public opinion,the recognition of hot topics and the analysis of hot topics.Topic Detection is the basis of research on hot topics in public opinion.Most researches focus on the improvement of topic clustering methods,ignoring the representation of topics.This paper combines single-pass incremental clustering and weighted association analysis to propose real-time topic detection and representation.Through the theoretical analysis and simulation experiment on small data set,the feasibility of this method is proved.The recognition and analysis of hot topics are the focus of research on hot topics of public opinion.Based on the evolution of hot topics,this paper divides hot topics into frequent and occasional and an analysis pattern with time constraints is constructed for frequent hot topics.The analysis of the hot topics is often only focus on the topic itself,rarely considering the relationship between different hot topics.Therefore,this paper proposes the temporal association analysis of hot topics and studies the relative concept and algorithm.Based on the real estate data from July 2015 to October 2015,this paper uses the time tag of the news corpus to simulate the network flow data,and uses the single-pass clustering to classify the public opinion information.Furthermore,we use weighted association rules to represent the topic,and choose one topic as an example to simulate the process of hot spot tracking,and find out the evolvement process of hot topics in different periods.Then,temporal association rules are extracted from the hot topic sequence of public opinion to quantify the temporal association among topics,which describes the driving effect and lag effect of public opinion topic.It further proves that this method is feasible in the analysis of public opinion hot topics.