Research On Hot Topics Discovery And Evolution Rule Of Weibo

Weibo is a broadcast social media and network platform for short information sharing,dissemination and acquisition based on user relationships.Users can access a variety of mobile terminals such as PCs and mobile phones,and use them to publish,comment,and forward mechanisms to realize instant sharing and dissemination of information in multimedia formats such as text,images,and videos.In recent years,with the rapid development of mobile communication technology and the increasing trend of mobilization of information acquisition methods,the number of user groups on the Weibo platform has expanded rapidly,content construction is increasingly enriched,with the speed of dissemination has grown geometrically,It not only brings people a new media experience,but also has become an important information release carrier at all levels of social life of netizens.It also plays an key role in public opinion as the influence continues to expand,and become an important source of current social hotspot issues and an important carrier for discussion and expansion.Especially for emergencies,with the characteristics like participation subject generalization,the three-dimensional distribution of communication,and the speed of dissemination,Weibo could expand the social effects.Therefore,it Caused widespread concern and be a significant academic problem about how to have a good command on the evolution rule of Weibo hot topics,and to supervise and manage public opinions effectively,also assist the government's management decision.According to the technical route of Weibo topic acquisition and preprocessing,which is topic discovery,heat assessment and evolution rule analysis,this research will focus on the extraction and evolution of Weibo hot topics.The main research work will include following four aspects:(1)Acquisition and preprocessing of Weibo informationA multi-machine parallel Weibo data acquisition method based on scrapy distributed data crawling technology framework will be stated.And Weibo advertisement information filtering method based on SVM algorithm and a special symbol processing mechanism will be proposed.A web crawler prototype system is developed.Through engineering practice,160,000 pieces of Weibo data can be obtained and apply to the subsequent research of the thesis.(2)Extension of microblog short text informationA short text information expansion strategy based on comments,and a comment screening principle based on the original microblog topic clusters and a screening mechanism,which solves the difficulties brought by text sparse text to the text analysis.Through experimental verification,the extended Weibo text information from this method will show higher accuracy in text clustering and topic extraction.(3)Weibo topic extraction and heat assessmentThis research will propose a Weibo topic extraction method based on text extension and dynamic topic model and a time series-based Weibo heat evaluation method.Based on the gensim framework provided by Python,by using the IdaSeqModel method,the DTM dynamic theme model and the heat map of the Weibo topic will be constructed,and compared with the manual labeling results,good experimental results will be obtained.(4)Analysis of the evolution rule of Weibo hot topicsThis research will propose a mining method of Weibo hot topic evolution rule.This method builds a co-occurrence network of evolutionary words based on time window,and divides the hot topic of Weibo into several communities according to the evolution word category.The evolution of hot topics will be explored by building community associations and correlation strength calculations in the time dimension.The example proves that this method can better mine the evolution rule of hot topic of Weibo,and can guide the evolution of hot topic of short text self-media to some extent.