Research On Hot Topic Classification And Heat Prediction Model Of Weibo

In recent years,with the development of social network media and the increasing number of Internet users,social media is becoming the source of the emergence and fermentation of hot social events.In the face of massive data,how to obtain effective information quickly and accurately has become a hot topic.Therefore,this paper studies the hot topic classification and the hot trend prediction technology,aiming at classifying the disordered topic information on social media,and predicting the trend of heat development for each topic category.Based on the analysis of the current situation of text classification and the principle that different parts of speech have different importance in semantic expression,a topic classification algorithm based on POSTLDA model is proposed.Divide the corpus to be divided into noun set,verb and adjective set,and other word sets according to their parts of speech.Set the weight coefficient to adjust the weight of different sets,The feature vector set of the topic set is established by using the document subject vector under three word sets.Then the clustering algorithm is used to realize the topic classification.The experimental results show that the text classification algorithm of the base POSTLDA model proposed in this paper is based on the algorithm of text classification.In this paper,we make full use of the different contribution degree of different parts of speech words to the classification result and obtain better classification effect.In view of Weibo's structural characteristics,the heat measurement formula is constructed by selecting the heat performance characteristics,and the topic heat prediction model based on PSO optimization BP neural network is proposed.The time series model of topic heat intensity is constructed as the input of the prediction model.According to the optimized prediction model,the prediction of topic calorific value is completed.The concept of heat growth rate is put forward to measure the change trend of topic heat.The calorific value of the next topic time slice is predicted by experiment.The experimental results show that the topic heat prediction based on PSO is optimized by BP neural network.Compared with the not optimized model,the model has higher prediction accuracy and can well simulate the change trend of topic heat,which has certain guiding significance to the reality.