Research On Modeling And Situation Analysis Theory Of Social Network Based On Attribute Graph

With the constant growth of internet, it has great significance to study the socialnetwork based on Web which has attracted more and more wide attention and anumber of research results have been made. However, the most researches on socialnetwork are based on the typical graph theory which ignores the attribute of node,edge and their relationship, so it could not well reflect the characteristic of dynamic,implicit fuzziness, information roughness and uncertainty and multidimensionalrelation of social network on the Web.In allusion to the complex features of socialnetwork, the theory foundation of modeling the social network is studied and theefficient method for analyzing the trend of the entity on the social and its structure,so as to achieve the purposes that can understand the state of the multi-dimensionaluncertain social networks accurately, and mining the social network efficiently.First, in allusion to the feature that individual and link have attributes in thesocial network, on the basis of traditional graph theory, it is built a new structure todescribe the complex social network—“attribute graph”. The basic property of theattribute graph is studied. On the basis of the attribute graph theory, the model ofrough attribute graph is built by integrating the rough set so as to describe theincomplete nodes and links relationship in the social network. Further more,considering the dynamic of nodes and links relationship in the social network, theS-rough attribute graph model is built by integrating the S-rough set theory. Therelationship between the S-rough attribute graph and rough attribute graph isdemonstrated.Secondly, it studies the sub graph matching problem in graph query and graphsearch of social networks based on the attribute graph.The decision algorithms areput forward. Based on the rough graph, the rough center area is defined and itsmining algorithm is designed which is proved efficient by an example.Based on theS-rough attribute graph, the relation between the transfer function and the graphroughness is proved so that there will have a simple and convenient method for analyzing the social network dynamically.Thirdly, On the basis of mathematical model constructed in the social network,considering the characteristic of multi-dimensional, uncertainty of nodes and theirrelationships in the social network, the model of set pair social network analysis isbuilt by applying the set pair analytical method. In view of the model, the conceptsof λ network central community and α relationship community are proposed.Further more, the mining algorithms of static and dynamic are designed which isdemonstrated the effectiveness and rationality by the experiments.Finally, the set pair potential is expanded into the generalized set pair potentialand the corresponding trend level table is built. The relation entropy is expanded intothe generalized relation entropy. The trends of the individual, relationship and thewhole social network are analyzed.In view of the generalized set pair potential, themethod is given how to calculate the network entity similarity based on attributesand relationships. It is demonstrated the method is reasonable by an example.