Research On Parallel Mining Algorithms Of User Relationship Strength For Mobile Social Network

In recent years,with the development of location and communication technology,the rapid rise of social networks,applications and media and location based services have spawned the mobile social networks,which provides a new perspective and idea for social network mining.Social network mining is not only a new and popular research field,but also a field of interdisciplinary research.User relationship strength mining is an important content of social network mining,which can be widely applied to user privacy protection,social recommendation,micro-blog sentiment analysis,network information dissemination research and public opinion monitoring.This paper studied the problem of user relationship in mobile social network,and proposed a parallel mining algorithm for user relationship strength in mobile social network.The innovations are as follows.1.Propose a calculation method UISFW for user interaction strength based on features weighting.UISFW extracts interactive features,which contains interaction behavior and interaction mode,and give different weights to them.Different weights reflect different influence of different features to interaction strength.2.Propose a parallel algorithm PAUFMP for mining user frequent moving patterns with time-constraints.PAUFMP finds user frequent moving patterns with sequential pattern,considering time and space factors,and suits for massive spatio-temporal data mining.3.Propose a parallel mining algorithm PMAURS for user relationship strength in mobile social networks.PMAURS combines user interaction data and trajectory data in mobile social network,mining relationship strength from the regularity of interaction patterns and frequent moving patterns.4.Propose user grouping methods based on user relationship strength,including EquiGroup,NormGroup and ExpoGroup.Which combine qualitative and quantitative analysis and discrete and continuous representation,reflecting social relationship and intimacy intensity between users.A series of related experiments show that the accuracy and ranking effect of the proposed algorithms are highly consistent with the real results.Meanwhile,user grouping methods can accurately reflect social relationship and intimacy intensity between users in different scenarios.The research on the mining algorithm of user relationship strength in mobile social networks is true and effective,and has its value.