Range Constrained Group Query On Attribute Social Graph

With the wide use of location-based mobile devices(such as smart phones),locationbased services and social networks have developed rapidly.People's demand for geo-social network has become more personalized,which promotes the diversification of query functions of geo-social network.In order to meet the personalized needs of users,this paper proposes the range constrained group query based on attribute social graph(RCGQ),which returns t user groups with the best keyword score that meet social constraints,spatial constraints and user group size constraints.Firstly,this paper proposes an extension method based on combination.In the process of expansion,only text information and user group size are considered.When the size of the user group meets the scale constraint,that is,after the expansion,it judges whether it meets the social constraint,and calculates the keyword score of the user group that meets the constraint.The t user groups that satisfy all constraints and have the best score are returned to the query user.Secondly,three social group expansion methods are proposed.The single-multi group expansion method adopts the group expansion method in the existing literature,and uses the breadth search method to expand the user group.The multi-multi group extension method and the single-single group extension method extend the user group in the way of deep search,and propose effective extension rules for them respectively,so as to avoid the repetition of the user group and ensure the completeness of the results.Return to t user groups that satisfy all constraints and have the best keyword score.Thirdly,in order to further improve the performance of the query algorithm,this paper discusses the reason why the vertex combination can not be the result of the query from two aspects of preprocessing and query process,and proposes an effective pruning strategy to make the query users quickly find the qualified user groups.Finally,experiments on real datasets are carried out to verify the effectiveness of the proposed algorithm.