Research On Social Networks With User Intimate Relationship

With the popularization and development of computer networks,social networks(SNs),e.g.,Facebook,Twitter and Google+,have become the most common platforms and tools for people's social activities,including interrelation,making new friends,sharing information and commodity recommendation.Under the rapid development of networks and mobile devices,the number of user grows day by day.Users' internal social attribute and the interrelation among users become more complicated.The relationship between users has become an extremely important reasearch topic in SNs.Meanwhile,with continuous emergence new social network applications,the choice of items for user is more affected by the increasing social information.How to give the efficient social recommendation scheme becomes a hotspot research.In this dissertation,we make deep research on hotspot issues of social network by the intimate relationship among users,and describe main contributions and creatives as follows:(1)Existing studies do not involve the special social network,which has uncertain number of adding social members and interval of members join follows exponential distribution.This type of social network exists in social activities widely,such as membership club.Therefore,this paper uses modified mean-field theory to study the degree distribution for this special social network.In order to increase precision of analysis,we propose an intimate relationship and spreading coefficient based algorithm,which determines the number of the joining members through the intimacy among members.The results of simulation show that the degree distribution of this social network follows power-law distribution and small-world characteristics.Moreover,we compared our algorithm with the existing algorithms,and our algorithm can reduce the time complexity significantly as well as achieving 29.04% higher precision and 42.69% lower computation time.(2)Existing approaches do not fully consider the relationship among social members,and with the problem of high computing complexity.We introduce the intimacy among nodes to detect community in social network.By reducing the degree of intimacy matrix between the communities,we obtain the accurate small community detection firstly.Then,in order to reduce the algorithm complexity,the intimacy-based algorithm for community merger was proposed to finally obtain the community detection.At last,through analyzing of theoretical and experimental,compared with the existing algorithms,our algorithm drops the time complexity.On average,our method is lower than the newly method with 39.06% and 47.09% in terms of iterations and computation time based on the precise community detection.(3)Due to uncertain network connectivity and complicated node intrinsic attribute,efficiently data forwarding in mobile social networks(MSNs)becomes challenge.To address the problem,this paper proposes an efficient geography intimacy based data forwarding scheme(GIDF)for MSNs to achieve h igher delivery ratio.In GIDF,we firstly detect several communities in MSNs by intimacy based dynamic community detection algorithm,which is proposed in the second work.Then,we propose a novel metric,named geography intimacy,which can quantify the node's geographical information and the friendships between nodes.Based on geography intimacy,we further propose a routing algorithm to forward data.Compared with the geography intimacy between nodes,the next hop is found,further find the route of data forwarding by doing the similar operations.Extensive simulations on real data with the ONE simulator show that GIDF is more efficient than the existing algorithms.With only 6.28% higher average latency,our algorithm can achieve 50.13% higher delivery ratio and 27.29% lower overhead ratio.(4)Since existing social recommendation methods ignore user's unreal ratings and do not fully understand the further social relationship between users,their recommendation accuracy tends to be significantly worse when the missing information rate is high.In this paper,we propose the intimacy among users to obtain a user-item objective rating matrix,which can avoid users' unreal ratings.For the sake of better predicting ratings,a user-item sub-block is presented,which to cluster intimate users and semblable items.Then,the sub-block can be detected through intimacy among users and similarity between items.In order to improve social recommendation accuracy,we propose a social contribution degree and social similarity based matrix factorization method to predict scores in sub-block.The final predicted ratings are obtained by combining all sub-blocks.Top-N items with highest predicted scores are recommended to each user.Systematic simulations on real world data set have demonstrated that our approach outperforms the state-of-the-art models by at least 14.59% and 10.5% in terms of Mean Average Precision(MAP)and Normalized Discounted Cumulative Gain(NDCG)respectively.In summary,we propose the concept of intimacy at first,and utilize spreading coefficient and mean-field theory to analyze degree distribution in special social network.Then,this paper proposes an intimacy-based dynamic community detection algorithm.On the basis of community detection,we study data forwarding in MSNs.At last,the intimacy among users is used to detect sub-block in user-item objective rating matrix to improve the accuracy of social recommendation systems.Our research enriches the methods of degree distribution,community detection,data forwarding in MSNs and social recommendation systems,and provids significative exploration for the relationship between users.