A Trust-based Social Relationship Mining System

In recent years,with the vigorous development of the Internet,more and more users communicate and interact frequently on the Internet,and a large number of social platforms have been established and developed rapidly.However,the huge amount of information and the huge number of users in the social platform make it impossible for users to quickly and accurately find the "friends" they are interested in,which greatly increases the cost of finding new "friends" and reduces the user experience of using the social platform.In order to solve this problem,social relationship mining system emerges as the times require.By using the information of users in social network,social relationship mining system can effectively predict the social relations among users in social network,and then recommend users who may be concerned to target users,so as to reduce the impact of information overload on users.Although recommendation algorithms have been widely used in social networks to mine the social relations among users,there are still many challenges in practical application,which are as follows:(1)in the process of mining social relations,the system often needs to call the user information many times,and frequently visit the user information,which is easy to disclose the user's privacy information,thus causing losses and negative effects to users,It greatly reduces the user's willingness to use the social platform,resulting in a large number of user loss.(2)The rapid development of social platform makes the social relationship between users become complex and diverse,and users want to find new "friends".However,the traditional social relationship mining system can not distinguish whether the relationship between users is "friend" or "enemy",which makes the recommendation result not necessarily accurate.Aiming at the above problems,this paper improves the traditional social relationship mining method,proposes a link prediction method based on simhash to mine the social relations and types among users,and applies the research results to the social relationship mining system.This paper mainly focuses on the following aspects:(1)Firstly,this paper uses the link prediction method to analyze the user's behavior to obtain the user's interest preference information,and then to predict the potential and unconnected social relations among users.Aiming at the problem of user privacy protection in social network,this paper applies simhash technology in link prediction method to map high-dimensional user rating information into low-dimensional Boolean feature vector,which effectively protects user's privacy information and improves user's experience of using social network.(2)Secondly,this paper uses the concept of "trust degree" [1] to distinguish the types of social relations.Through fuzzy calculation model,the trust degree between users is calculated according to the user rating information.When the trust degree is positive,the relationship between users is trust,that is,"friend",otherwise,it is the relationship of distrust,that is,"enemy".Finally,users with trust relationship are recommended to the target users.In order to verify the feasibility and effectiveness of the improved method,this paper designs and implements a comparative experiment based on the real epinions data set in the real world,and finally proves that the method is feasible and has advantages.(3)Finally,the development of social relationship mining system is completed.On the research results of(1)and(2),based on the B / S architecture model,according to the analysis of the system feasibility and user needs,this paper designs and implements a social relations mining system.In this paper,we use python programming language and Django development framework to implement the various functional modules of the social relationship mining system under pychar development environment.Then,we use the link prediction method based on simhash to mine new social relations and calculate the trust between users,so that the social relationship mining system can more timely and accurately recommend positive trust to users To provide users with a good experience.