Research On Dynamic Profile Generation Of Microblog Users Based On Interest Transfer

With the rapid development of Internet technology,social media,represented by microblog and wechat,has won people’s high dependence by virtue of its unique advantages of popularization,diversification and free participation.The huge scale of users and the interaction between users make social media produce massive information data all the time,which provides an important data source for the construction of user profiles.User profile is a user tagging representation based on real data.User profile is of great value for accurately identifying user needs and interests,and has been widely used in personalized recommendation,precision marketing and other aspects.Microblog as a representative social media in the era of Web2.0,the construction of its user profile has gradually become a research hotspot.In this paper,on the basis of the existing research on user profile,aiming at the problems of incomplete and inaccurate user feature extraction in the existing methods of microblog user profile construction,a microblog user dynamic profile construction method based on interest transfer is proposed.In this paper,the dynamic profile of microblog users includes two parts:static attribute tags and dynamic interest tags.The static attribute tags are extracted from the basic information of users’ microblog platform home page;the dynamic interest tags are used to mine the dynamic interests according to the interactive microblog set between users.The main work of this paper is as follows:(1)User static attribute tags extraction: Obtain user related attribute information from user’s microblog home page personal information,including gender,age,region,school,company,personal profile,microblog authentication and self checking tags,process the original text,determine the tag weight by using analytic hierarchy process,and generate static attribute tags set.(2)Dynamic interest tags generation and update based on interest transfer: According to the analytic hierarchy process to determine the weight of different interest behavior reflecting the degree of user interest,and use it to revise the weight of interest words to obtain the initial set of user interest words;according to the life cycle theory to describe the user interest behavior cycle curve,construct the time attenuation function of interest transfer,so as to realize the dynamic update of user interest word set,that is,the dynamic generation and update of interest tags in user dynamic interest tags set.Finally,the static attribute tags set and the dynamic interest tags set are integrated to construct the dynamic profile of microblog users.In order to verify the effectiveness of interest transfer based microblog user dynamic profiles,the performance of different user profiles for microblog personalized recommendation is compared on real microblog experimental data sets.The experimental results show that:compared with the existing methods of microblog user profile construction,the microblog user dynamic profile based on interest transfer in personalized recommendation has better @6)and better performance.