Research On Multi-dimensional Topic Interest Mining Technology For Microblog Users

As a social networking platform for sharing short and real-time information through attention mechanism,micro-blog has attracted more and more users.By 2018,the number of registered micro-blog users has exceeded 700 million,and users publish hundreds of millions of Posts every day.Massive micro-blog data provides a wealth of personal information sources.In the micro-blog platform,users can publish and share information about daily life,news events and other interesting information,which contains users' interest information,and is an important data source for mining micro-blog users' interest topics.Micro-blog users form a social circle by paying attention to other ways,and the content of micro-blog published by users is transmitted in real time by forwarding,which makes micro-blog become an important social medium for users to obtain the latest information.The interest mining of micro-blog users can help users obtain the information they need from the massive micro-blog data,and provide personalized recommendation services for users.It is a hot research topic at present.In order to more comprehensively and accurately explore the interest topics of micro-blog users,and comprehensively consider the micro-blog content,such as microblog users' original,forwarding,likes and comments,as well as background information,this paper proposes a multi-dimensional interest of micro-blog users.The method of modeling,which excavates the professional interests,hobbies and multi-level interest of micro-blog users from the breadth and depth.Firstly,on the basis of LDA(Latent Dirichlet Allocation)topic model,introducing some prior knowledge(PK: Prior Knowledge)such as introduction information,authentication information and semantic knowledge of vocabulary,an improved PK-LDA model is proposed to mine more accurately the interest topics of micro-blog users;secondly,the interest categories of users,i.e.user specificity,are mined through interest fusion in breadth.In order to more accurately identify the interest categories of micro-blog users,we construct a user interest database and get more accurate and reasonable user interest categories.The experiments performed on the dataset of a large number of real micro-blog user interest information collected by the Python crawler program verify the validity of multidimensional interest modeling.The experimental results show that the interest fusion and hierarchical clustering algorithm can accurately mine the breadth and depth interest of micro-blog users,that is,the user's professional interests,hobbies and multi-level interest topics of users' multi-level interest;and LDA topic model.In contrast,the improved PKLDA topic model can more accurately and effectively mine the interest topics of microblog users;In the aspect of user's wide interest mining,the accuracy of interest topic mining after interest fusion is higher;building interest database can more accurately and reasonably identify the interest categories of micro-blog users.