Research On The Model Of Social Network

Internet is a huge and complex artificial system integrated from all human wisdom and resources which makes it difficult to be separated from human’s economies, cultures and lives, and therefore how to protect the positive and healthy development of the Internet ecosystem is an important thing for the whole mankind.Traditional network security methods like firewall, data fusion method has encountered a problem. It is a new approach to study social network as an open complex giant systems, which is totally different from Reductionism, a priori model and doll model.The main contents of this paper include:1. Study the traditional Internet models, points out their problems, and then put forward the "tomographic model", which extracts social network tomography from the flow parameters, flow diagram and the application layer;2. Study mainstream mathematical algorithms, and their applications on the extractions of key stream parameters, flow diagram structure, social network community structure and hot topic;3. Put forward the basic framework of social network behavior analysis system based on "tomographic model" and the realize of the two key technologies of the system, which are dual pipelining regular expression matching based on FPGA and mathematical algorithms parallel acceleration based on GPU.The first chapter is an introduction, following a brief history of the Internet, and then find that the Internet has entered the era of social network, finally elaborate the limitation of conventional research method, and the research necessity of Internet model.The second chapter introduces the basic concepts and principles of complex network, analyses the existing Internet network models (topological model and HOT model), and points out the existing problems of the models.The third chapter introduces the theory of open complex giant system, human behavior dynamics researches, as well as mathematical tools needed, as a basis of Internet "tomographic model"In the fourth chapter,"tomographic model" is proposed based on the theory of open complex giant system,"tomographic model" is a data-driven model based on people’s behavior on the Internet and behavior can be perceived by stream clustering. Stream information, extracted by mathematical methods, can tell the people’s behavior, which is our interest and purpose.The fifth chapter is the analysis and interpretation of the Internet mail data, as an example of "tomographic model"The sixth chapter proposes the basic framework of social network behavior analysis system based on "tomographic model", and the realization of the system key technologies and algorithms in FPGA and GPU, to meet the high bandwidth and real-time requirements in the backbone network.Chapter7summarizes the main ideas and the innovations in this dissertation, and points out the direction of future work.