Research On Friend Recommendation In Mobile Network

Thanks to the big success of social network in the past decade, connecting with friends through SNS has become part of the fabric of everyday life. As rapidly growing mobile technologies begin to take over the world, the social network services have been adapted to mobile environment. Playing a great role in SNS, new friend recommendation feature could discover potential social connections of users. It would have great significance on user preference detection and personalized service recommendation. The main task of mobile friend recommendation service is to improve the accuracy as well as user satisfaction by utilizing mobile context, mobile social network and other relevant information.In this paper, a context-aware friend recommendation algorithm based on mobile phone data is proposed to provide personalized friend recommendation service in mobile network. Firstly, the communication data of mobile device is used to calculate the communication social intensity. After that, data collected by short-range technology is applied to seize the geographical social intensity between users. Finally, on the basis of the results got from previous procedures, the method uses a graph-based algorithm to obtain the friend recommendation lists. The experimental results reveal that the method improves the accuracy and performance of friend recommendation service.Unlike the social networks center on social graphs, interest-based social networks focus on user interest and user preference. Therefore, an interest-based mobile friend recommendation approach is proposed in this paper. The proposed approach constructs the user model by considering three aspects:user tags, follow relationship and micro-blog contents. User tags can be utilized to mine user preference. Follow relationship can be used to infer the user attention information. Micro-blog contents can be utilized to discover users’ topics of interest. After building the user model, the interest similarity value is calculated. And based on which the friend recommendation lists are generated. The design and implementation of the interest-based friend recommendation system are also presented in this paper. This application provides a variety of functionality for users to detect and interact with other peers in proximity. The recommendation lists are generated based on the interest similarity between users. Therefore it can help users expand the social relations.