Research And Implementation Of Community Detection Based On Geographical Feature For Social Networks

With the widely popularity of smart phones embedded positioning chips, promoting thedevelopment of traditional social networks to location-based social networks.Location-based social network which has the board prospects is the outcome of thecombination of location-based service, mobile internet and social network service.User will produce large amount of data containing geographical information, when usethe location-based social network. How to use the large amount of data which is generatedby users and contains geographical information, to analyze users’ behavior patterns,mobility patterns and geographical-aware community structure has become a hot topic.In this paper, we crawled data set from Sina Weibo. Firstly, we analyze thecharacteristics of the users’ geographical information, and then use the geographicalfeatures to calculate users’ similarity and to detect the overlapping communities. Finally, weproposed geographical-based features users’ similarity calculation algorithm andoverlapping communities’ detection algorithm, and we also designed and implemented ageographical-based overlapping community visualization tool.Studythe methodsof similarity calculation on the location-based social network. Byanalyzing the characteristics of the user data comprising location, we proposed a similaritycalculation method based on geographical features. We verify the effectiveness of themethod by experiments using the data crawled from Sina Weibo.Study the methodsof overlapping communities detection algorithm on thelocation-based social network, analyze the geographical features and the relationshipbetween users and improve the edge-clustering algorithm. On this basis, we design andimplement an overlapping communities detection algorithm based on users’ geographicalfeatures. Finally, we verify the effectiveness of the method by experiments.In this paper, we design and implement a geographical-based overlapping communityvisualization tool. This tool is divided into three layers: data layer, core layer and view layer.Data layer is to package and preprocess the data of uses’ check-ins. View layer is to display and trigger the corresponding events. The mainly of core layer is the implementation ofalgorithms, including similarity calculation algorithms and overlapping communitiesdetection algorithms. It is also responsible for handling corresponding events from viewlayer.