Core Node Mining And Group Recommendation Based On Online Community

With the rapid development of information science and technology,making the internet has been widely used,network information has shown high speed growth.Faced with massive resources of network information,how to quickly and efficiently obtain the valid information become research priority.The appearing of recommendation system provides a solution to this problem.Recommendation system can recommend information that may be of interest to users,greatly reducing the cost of users searching for information,while businesses can use this tool to achieve precision marketing.With the development of technology,online communities and social networks are gradually on the rise,through the network,human to human interaction is becoming more convenient and more frequent,more activities have been expanded for group.Organize group activities have become more common,thus bringing target group recommendation requirements.Previous relevant studies focused on personal interest,in contrast,the research about group recommendation is still in infancy.And the Group recommendation system is different from the simple sum of the individual recommendation.There are preferences disagreement factor among members of the groups.It is still need an in-depth study on somel aspects of the group recommendation.Online communities have their own unique characteristics.Online social networks is tending to be interest communities,users with similar interests independently formed small groups,the groups has distinct boundaries and often in groups.There is always exists core users in the group,and the core group has an important influence on group preference.Based on current research situation of the group recommendation,considering the features of online communities and the influence of core groups on group recommendation.In this paper,proposes a new group recommendation method based on the preference of the core group.Firstly,this paper uses social network analysis algorithm to analysis the online communities.It uses core-node degree to measure the core level of the nodes based on the point centrality degree,by considering the centrality in the network,the center coverage factor,the user's properties and the importance of fans.To spot the core nodes based on the qualitative analysis.More in line with the practical application of online communities.Secondly,due to the dominance of the core group in group,and its important impact on the overall group preference,on the basis of previous studies,this paper determines the core group from the identified core nodes,uses the average satisfaction strategy to model core group preference,and get the initial group preference model.Thirdly,considering the disagreement factor between non-core group users and core group,it uses the amended core group preference model as the group preference model for group recommendationFinally,it uses the data of Douban movie interest groups to experiment.The results show that this method is effective.