Research And Application Of Social Relationship Based On Social Networks

Online Social Network (Online Social Network, OSN) service, such as Facebook, Microblog, has enjoyed great popularity on the Internet with a rapid increase of users, which become indispensible for socializing and acquiring information.User and their relationship are the two basic elements of constructing and maintaining the online social network. In this paper, we focused on user relationship and have researched strengh, type and dynamics of relationship. This research could provide theoretical foundation for related issues in OSN such as recommendation, privacy preserving, etc., thus service of practical value could be offered by OSN operator.In this paper, we surveyed theories and achievement of user relationship in some areas and did comparison of these algorithms, based on which we propose our research framework about user relationship in OSN. Multivariate Stepwise Linear Regression (MSLR) is presented to calculate relationship strength and Random-Walk based Relationship Type (RW-RT) is used to identify user relationship type by taking use of user profile information, user generated content and topological structure of users in OSN. Data from Sina Microblog are used to verify the validation and effectiveness of these algorithms. The precision of MSLR is about80%and the accuracy of RW-RT is about85%. Besides, for the issue of dynamics of relationship, we start with the temporal pattern of user behavior and Wavelet Transformation and Dynamic time warping for K-Medoids (WT-DKM) is proposed to summarize the typical temporal pattern of microblog users, which explains that dynamic behavior is the key aspect that leads to the variation of user relationship.In addition, an application inserted in Sina Microblog is developed based on the algorithms stated above, which provides the functions of measuring relationship strength and managing friends by automatic grouping. This application also implicates the practical significance of our study.