Research On Social Media-oriented User Recommendation Method

With the development of the society,the social media is diversified and complicated.The number of users is also growing rapidly.The main purpose of using social media is to make friends,to maintain the relationship between friends,and to expand their social scope.However,social media has a large scale of users and huge amounts of datainformation.It's difficult for social media users to find the objects and information they are interested in.Therefore,it is of great significance to study how to recommend users in social media.Recent years,user recommendation method has become a hot research in social media,but there are still some problems in these user recommendation methods.First of all,user information available in existing user recommendation methods is relatively little.These methods are basically used tags and user information,which can not accurately determine the user's interest.Secondly,existing user recommendation methods search the entire cyberspace.The time and space cost is very large the results is not precise because the cyberspace is generally large.Aimed to the above problems,the paper mainly studies a kind of user recommendation method based on social media.The main work of the paper includes the following aspects.Firstly,the paper analyzes the problems and research results of current user recommendation method.Based on these,the paper gives the formal description,related concepts and terminology of user recommendation for basic social media.Secondly,a social media-oriented user recommendation Method is presented.Then the basic idea and formal description of the algorithm are given.The complexity of the algorithm is analyzed and discussed.This method includes three key steps: method for constructing user candidate set based on Re-Posting Network(forwarding network),user interest extraction method and Top-k user sorting method.Aiming at the Re-Posting Network,a N layer forwarding network model is proposed to discover the users set with the potential same interest.On the basis of N layer forwarding network,the paper propose the user's topic interest extraction algorithm based on topic model and forwarding network model.We can obtain the distribution of user's topic interest by this algorithm.Based on the distribution of user's topic interest,different weight are used to distinguish between the forwarding interest and the transmission interest,and then Top-k user sorting method is proposed to find the best K users.The basic idea,the formal language description and the evaluation indexes of the algorithm involved in the above three parts are elaborated in the paper.Finally,experiments are designed and implemented to verify the feasibility and effectiveness of the above algorithms.We pre-handle the data extracting from parts of Twitter data set and DBLP data set and then extract the experimental data meet the requirements of the paper.Then the experiments are designed and implemented to verify the feasibility and efficiency of the algorithms mentioned above.The algorithm is analyzed and compared in terms of the number of subjects,the quality of the keywords,the running time of the algorithm,and the scalability of the algorithm.The results prove the feasibility and correctness of the method.