Research And Realization Of Semi-automatic Friend Recommendation Mechanism Based On Realistic World Evidence In Social Network

In recent years,with the large-scale expansion of the Internet and the popularity of a variety of virtual social networks,SinaWeibo,Tencent QQ,WeChat,Twitter and other virtual social network widely get into people's real life,people shares their own information and living states each other in real time.For this type of social network,users are connected to each other through a "attention-focused" approach.Because this kind of social network is non-real name system,we often cannot identify the "person of concern" or "fan" status,and with the increase of information,the process to determine becomes more and more difficult.However,these users in the real world is that we may also be familiar with or well-known friends or classmates.separation between the social network and the real world network of people is arisen,and in fact,the social network is an extension of the reality network.Therefore,it is particularly important to help users find the real world of friends in a huge social network.The characteristics of the social network and the real world network is analyzed in this paper,and the semi-automatic evidence is used to connect the social network world with the real world network,and the proposed mechanism is tested on the Sina cloud.The specific work is as follows:(1)The model of social network and interpersonal relationship is studied,and the recommendation mechanism and the related achievements are analyzed in this thesis.(2)Based on the research of the filtering algorithm and user's similarity measure method,the results of similarity measurement based on five different attributes are tested and analyzed,and similarity measurement of the integrated five different attributes is used and tested in this thesis.The results show that the measurement results of the integrated attributes can significantly improve the accuracy,and with the increase of data size,the performance is more stable.(3)In order to find the friend relationship between the real world and the virtual network,a semi automatic recommendation mechanism is proposed in this thesis and it is applied into the Sina micro-blog system.Firstly,the user's potential friend in the virtual network is obtained and classified,and then,the evidence is provided to determine their identity in the real world.If the system has the real identity data of the friend,the corresponding relationship is completed.If the system does not have the relevant data of the friend,the identity of the friends is determined with the corresponding evidence that is obtained through the whole micro-blogging,and is stored into the database of friends.The relevant evidence is inputted to find friends,and for each friends recommended,the relationship between the user and the friend is presented.Finally,the recommended effect and performance of the system are analyzed.