Research On Weibo User Interest Model Based On Topic Words

Nowadays,the development of mobile internet in our country has gradually matured and stabilized.The development of the industry pays more attention to the content quality,model innovation and platformization trend.The development direction of the advertising market is mobilization,socialization,video,which makes the commercial value of Weibo more and more recognized by the industry,becoming a social platform for advertisers in the industry to compete to purchase.Based on the user data of the Weibo platform,this article builds a Weibo user interest model.First of all,through reference reading of historical documents,only original microblogs are used as data sources,and it is not accurate enough to mine user interest preferences.Statistical analysis of a series of behavioral data of Weibo users,and ultimately determine the source of the data for this article is the user original microblog,user focus grouping and custom tags.For different data sources,different processing methods are used to preprocess the data,and then a microblog user interest model is constructed through corresponding rule methods.Finally,three basic models are obtained,and these three parts are effectively combined by an algorithm,and finally the micro Bo user interest model system.The contribution of this paper is to integrate the user interest model obtained through the three ways of Weibo user original microblog,user attention grouping and self-defined tag,and more accurately mine user interest preferences.On the basis of previous LDA improvement,the model was applied to the in-depth mining of Weibo user interests,and the prior probability of the model was replaced with a posterior probability to determine the user interest model,which improved the calculation efficiency of the topic model and ensured The accuracy of topic interest mining.Finally,the model is evaluated through evaluation indexes such as recall rate and precision rate,and the applicability of the model applied to personalized advertisement recommendation is analyzed.