Research And Implementation Of Privacy-Preserving Friend Recommendation In Online Social Network

With the development of online social network,it brings a lot of innovative services for users,such as sharing your favorite pictures or videos,tagging your interested blog or users,initiating activities or appointment on the web.A variety of online social network services are increasingly similar to people's real life activities.Like real world social networks,the cornerstone of online social network is the friendship between users.Friend-recommended service is proposed to help users expand their social circle in online social network.However,a lot of friend-recommended service process have exposed the problem of user privacy leak,such as identity privacy and social relationships privacy leak.And the data privacy leak issues hinder the popularity and promotion of online social networks.To protect user privacy,user data need to be encrypted or anonymous before outsource to the cloud server(Cloud Service Provider,CSP).But,it is a challenging problem that effectively run friend recommendation under the encrypted and anonymous user data.This paper focuses on privacy-preserving friend-recommendation in online social network.The main contribution of this paper are:(1)For the problem of user identify privacy and social relationships privacy leakage on friend-recommendation process.We propose a privacy-preserving friend-recommendation scheme(K-degree anonymous friend-recommendation scheme,KFRS),It can effectively hide the user's identity privacy and social relationships privacy.Firstly,based on hyper-graph model to abstract the topology of online social network,and existing K anonymous technical,we present a segmentation algorithm for anonymous user data,and this algorithm can hide user identity privacy and the social relationships privacy.Secondly,based on the phenomenon that peoples who have more same interests are more likely to become friends,we design a similarity calculation algorithm between users,and the friend recommendation is based on the similarity between users.Lastly,consider the encrypted and anonymous user data the encrypted and anonymous user data will decrease the utilization of friend-recommendation,we propose a segmentation tree algorithm to eliminate the impact which caused by anonymous algorithm.Through experimental analysis,we find that recommended scheme can ensure effectiveness while protecting user's identity privacy and social relationships privacy are not leaking,and highly scalable for a huge amount of users online social network.(2)Demand for the proposed privacy-preserving friend-recommendation in online social network,we construct a prototype of online social network,and achieve the basic social functions and user privacy protection aims.We construct the framework of online social network in CSP and develop the client in mobile platform.In server,we design and construct the database,and program the application programming interface.Also based on the encryption and anonymity scheme between server and client,we realize data privacy-preserving on friend-recommendation.In client,based on iOS mobile platform we develop user interface,and based on employ the application programming interface we realize the queries and stored of user data.Also based on the exchange of user data,we realize the basic social function.For the purpose of user privacy protection requirements when user data outsourcing to the CSP,we develop the function of anonymity and encryption of user data in client.