【作者】 王宏；

【导师】 徐志明；

【作者基本信息】 哈尔滨工业大学 ， 计算机科学与技术， 2019， 硕士

【摘要】 近年来,伴随着各种社交软件的不断出现,人们沟通信息的方式逐渐从线下活动转移为线上交流。由于信息沟通方式逐渐变得简便容易,人们的交友圈不断地扩大,社会网络随之变得庞大而复杂,通过社交网络认识志趣相投的人获得新鲜有趣的信息愈加困难,因此,基于在线社交网络的人物推荐算法的研究变的至关重要。通过推荐算法推荐能给网络用户带来更多新鲜信息的朋友,使社交网络信息更多样化,是目前社交网络研究的重点内容。经典人物推荐算法更多的是根据用户之间的相似性进行推荐,给用户推荐相似的好友,由于未考虑用户对新鲜有趣的异质信息获取的需求,从而给用户造成了一定的信息冗余。本文针对该问题进行研究,提出了社会网络中的弱关系人物推荐算法,推荐网络中和用户联系为弱关系的节点,给用户带来更多样化的异质信息,进而促进整体网络的信息流通。本文首先简要阐述了社会网络中对强关系和弱关系的定义,利用社区划分算法识别强弱关系,通过经典人物推荐算法验证了弱关系对于社会网络信息流通的重要性。在弱关系能够加速网络异质信息流动的研究基础上,提出在线社会网络上的弱关系人物推荐算法,开展对比实验,验证了弱关系人物推荐算法能够加强网络的信息流通,提出的弱关系人物推荐算法对网络异质信息的获取更为有效。同时在本课题研究过程中发现社区发现算法存在社区划分不均匀的问题,对弱关系的识别和节点的重要性计算造成了不利影响,针对该问题本文提出了一种新的基于“捷径”的弱关系识别方法,并结合网络拓扑结构给出了新的节点重要性计算方法,提出了社会网络中基于“捷径”和网络拓扑结构的弱关系人物推荐算法,进一步的提升基于弱关系的人物推荐算法的异质信息获取能力。更多还原

【Abstract】 In recent years,along with the emergence of various social software,the way people communicate information has gradually shifted from offline activities to online communication.As information communication methods become easier and easier,people’s circle of friends continues to expand,and social networks become large and complex.It is increasingly difficult to get interesting information through social networks.Therefore,research on character recommendation algorithms based on online social networks has become crucial.It is the focus of social network research by recommending algorithms to recommend friends who can bring more fresh information to network users and make social network information more diverse.The classic character recommendation algorithm is more based on the similarity between users,recommending similar friends to the user,and does not consider the user’s need for fresh and interesting heterogeneous information acquisition,thus causing certain information redundancy for the user.This paper studies the problem and proposes a weak tie recommendation algorithm in the social network.It recommends the nodes in the network that are weakly related to the user,which brings more diverse heterogeneous information to the user and promotes the information of the whole network circulation.This paper firstly expounds the definition of strong relationship and weak relationship in social network,uses community partitioning algorithm to identify strong and weak tie,and verifies the importance of weak tie to social network information circulation through classical character recommendation algorithm.On the basis of the research that weak tie can accelerate the flow of heterogeneous information on the network,the recommendation algorithm of weak relationship on online social network is proposed,and comparative experiments are carried out to verify that the recommendation algorithm based on weak tie can strengthen the information circulation of the network.The recommendation algorithm is more effective for obtaining heterogeneous information on the network.At the same time,in the research process of this subject,it is found that the community discovery algorithm has the problem of uneven community division,which has adversely affected the identification of weak tie and the importance calculation of nodes.This paper proposes a new "shortcut" based on this problem.The weak tie identification method,combined with the network topology structure,gives a new method of node importance calculation,and proposes a weak tie person recommendation algorithm based on "shortcut" and network topology in social network,which further enhances the recommendation of people based on weak tie.The heterogeneous information acquisition capability of the algorithm.更多还原