Research On The Method Of Social Networks Group Recommendation Based On Social Selection And Social Influence Mechanism

With the increasing application of web2.0 and development of Microblog,WeChat,PostBar and other types of social media,people communicate through various network applications.By doing so,they cluster into large or small communities.Interactions between community members frequently happened,and discussion a wide variety of topics.The traditional E-commerce company move forward to social business model,and add the function of social communication into trading platform.The group recommendation system for online social networks has become a hot topic in the research field of personalized recommendation.Different types of social networks in the macro indicators are similar,but micro-formation mechanism are not the same.If we can understand the formation of these networks and the features of the networks will help recommend the content users interested in,thereby enhancing the stickiness ofusers' and carry out social marketing.So the research has a high theoretical and practical value.Study shows that the formation mechanism of social network and the links between individuals can be divided into social selection,social influence and some exogenous effects.The formation mechanisms and features of diverse types of networks have big difference,it has brought challenges for the group recommended applications.The difference between group recommendation system and individual recommendation system is that the final decision is not be made by an individual,but it is the result of joint decision-making among group members.Therefore,through the study of social selection and social influence mechanism,implement social selection and social influence in obtain of user preference for group recommendation,group recommendation community identification and recommended strategy,will achieve good results.This dissertation first analyzes the microcosmic mechanism and features of networks,and then studies the implementation technology for social network group recommendation.Specific research work and innovation as follows:(1)Individual users in social network to establish links from the theory to explain and understand the role of social selection and social influence is easily.But in the real network,the two mechanisms are intertwined and promoted,how to distinguish between the two mechanisms,still lack effective methods and means.From analysis of the indicators of the real network,this dissertation determines the basis of quantifying social selection and social influence.And then for hybrid dual-mode social networks to establish social-affiliation network model,based on the principle of triangles closure simulation analysis,to study individuals join in networks and participation in community activities pattern.The results show that there are significant differences in the in-degree distribution,network structure,network community division,propagation characteristics and node influence,but there is no significant difference in the activity of the group under the different conditions of social selection and social influence effect.The simulation model is strong in explanation,and is in accordance with the actual reality of the network and the existing research conclusion,which shows that the simulation modeling of the two mechanism is reasonable.(2)Research on methods for obtaining preferences of user in social networks based on social selection and social influence.In order to realize the obtaining of individual preference for group recommendation system,we studied using the LDA model is used to represent the user's preference,and three methods of user preference are proposed.Second,from users' social selection behavior,according to users joint participate in the discussion of topics to find their nearest similar neighbors,we got users' preference from the nearest neighbors.Third is considering the user influence of nearest neighbors to further optimize users' preference.The experiment showed that the method of considering social selection and social influence is the best in our research.This brings a new idea to modeling social network user preferences.(3)Research on social networks community detection based on multi-dimensional social selection and social influence.In the real world,social network is often multidimensional due to the interaction of social selection and social influence,and there are many kinds of interaction between user nodes,so the traditional community detection method cannot apply this situation.This dissertation take into account the directly or indirectly,directed or undirected and weighted interaction links between users under the environment of social networks.And analyzes the multi-dimensional social relation networks construction method from two perspectives of social selection and social influence mechanism.We focus on the dimensions reduction of multi-dimensional social networks for community detection.The experiment also further introduces the user attribute information.Finally,the existing methods are used to identify the community on the network after the dimension reduction.The experimental results show that the method proposed is more in line with the real situation of the social network.The evaluation criteria of interaction degree and accuracy rate are more effective.(4)Group recommendation strategy of social networks based on social selection and social influence.In view of the situation that the social network user preference information is difficult to obtain directly.And the user's motivation to participate in the community is different,it is difficult to individual preferences aggregate into groups of preferences.So,this dissertation models the preference of group users on the basis of implicit feedback method.Then,aiming at the social network which will have different influence degree on the social selection effect and social influence effect,we judge social selection and influence on user preferences to realize the classification of users and groups.Finally,we discusses the generation process of group recommendation candidate scheme of two categories communities,and recommendation strategies is put forward.The experiment show that the proposed group recommendation strategy based on social selection and social influence,solve the group recommendation results coordination problem,and better than the general personalized recommendationmethods.In addition,it can guide the application of other social marketing.Our method with good scalability and practical value is higher.This dissertation analyzes the mechanism of social selection and social influence,as well as the features of the network from the micro perspective.And then applies the selection and influence of the two mechanisms to the group recommendation of individual users' preference,community identification and recommendation strategy.The research is helpful to understand the social network formation mechanism,structure function and group behavior pattern of the social network community,which is of reference value for the enterprise to construct the group recommendation system and the social marketing activities in social networks.