Research On Hot Spot Discovery And Trend Forecast Of Microblog Public Opinion

With the constant improvement of network facilities in China,the rapid development of mobile applications technology,microblog has from the initial meet the social demand of information communication platform gradually became popular opinion.The characteristics of the microblog public opinion that the sudden and fission information dissemination make it become a quickly reflection form of social public opinion,and to a certain extent,to lead the public opinion events.The influence of microblog public opinion on society in various fields is growing,but due to the particularity of its text,and affected by many factors present a nonlinear complex changes.Therefore,it has become a meaningful research topic for academia to efficiently find and extract valuable hotspot information,fleetly and accurately predict its development trend from microblog data.Thesis focus on the hot spot discovery and trend forecasting of microblog public opinion,and propose a method of microblog public opinion hot spot based on Biterm Topic Model.Firstly,to adapt the characteristic of short texts in microblog,we deal with microblog texts by the BTM,and improve the TF-IDF weight calculated algorithm,which is combined with the result of BTM modeling to vectorize microblog texts,considering the probability distribution of document merged with the semantic distribution.Which solves the problem of high dimension and sparseness of traditional model in text modeling,and then uses K-means clustering method to find hot topics.In the topic of development trends forecast,this article uses the total number of posts of microblog topic as a measure of the development trend of the indicators.Considering the complexity and non-linearity of topic development,fuzzy neural network is used to predict the development trend of microblogging topic.And the improved Particle Swarm Optimization(PSO)algorithm is used to optimize the parameters of fuzzy neural network.PSO optimization algorithm has good performance in global optimization and fast convergence.Fuzzy neural network in dealing with the complicated problems such as nonlinear,fuzziness has great advantages.By combining with improved PSO algorithm,fuzzy neural network can better play the performance of fuzzy neural network,and effectively solve the problem of microblog public opinion trend prediction algorithm of slow to converge,easy to fall into local optimal problem.Through the contrast experiments of Sina microblog data set,the validity of the method suggested in this paper is proved in the microblog public opinion discovery and trend forecasting.This method can solve problems of higher dimension and sparse property in text modeling of traditional model,and improve the quality of discovering hot topics.Effectively solve problems of complex model parameters and easily to run into partial optimization in microblog public opinion trend prediction,and improve the accuracy of predicting the trend of microblog public opinion.