Research On Community Detection Techniques Of Telecom Network Based On Users Communication Behavior Analysis

With the development of Telecom Network business, more and more users utilize TelecomNetwork. Users carry out various social communication activities by means of Telecom Network.The social characteristics of the user emerge through information exchange in Telecom Network.Acording to the characteristics of the situation in the process of information exchange, users areeasy to form various types of interpersonal groups. Various interpersonal groups can easily leadto a mass network effect.This dissertation is supported by the National High-Tech Research and DevelopmentProgram of China (863):“user relationships and behavior analysis technology in the publiccommunications network”. The article is based on the analysis of the Telecom Network usercommunication behavior, so as to realize the user community detection, to provide technicalsupport for the advancement of Telecom Network business and regulatory. The task focus on thetwo key technologies: First, the analysis techniques of user communication behavior in TelecomNetwork, using the technology to confirm the effectiveness of user communication behaviorcharacteristics to community detection. Second is community detection technique, the use of thetechnology to achieve community detection of Telecom Network. On this basis, combined withthe needs of the project and other mature technology, we realize the design of communitydetection subsystem in Telecom Network.The main innovations and achievements of this dissertation are as follows:The analysis and generation method of Telecom Network communtication behaviorcharacteristics is proposed. Aiming at the shortcoming of the user behavior analysis of theexisting, the method of combining classification process and social network analysis is used.First the paper analyse the distribution of the user various types of characteristics. On basis ofthis, this paper proposes the method of setting the characteristics weighting, and then combineswith the improved C4.5decision tree algorithm for user classification. Second, based oncategories of users, the paper combines the connection data of Telecom Network to generate thenetwork link feature, which can profoundly reflect the social relationships between users. Theabove work provides the main basis of communication detection.A community detection algorithm that combines user characteristics and network linkcharacteristics is proposed. In the design of the algorithm, based on the business structurecharacteristics of Telecom Network, The node communication characteristics, and structuralcharacteristics is considered, combining with network link characteristic feature, which canprofoundly reflect the user’s social relationships. Aiming at large data features of Telecom Network, aggregation algorithm is used to achieve the detection of the community, since lowcomplexity of aggregation algorithm. Experimental results show that the proposed communitydetection algorithm is superior to the melting the two characteristics algorithm and the users canbe allocated to the corresponding community quickly and efficiently.Based on the above technology and the needs of the project and the existing maturetechnology, Telecom Network community detection subsystem is designed. The subsystemadopts hierarchical structure, and consists of data access layer, data interface layer and businessprocess layer. It process call communication data, by means of communication behavior analysisand communication detection algorithm. Tests show that the performance of the subsystem meetsthe needs of the project.