Reaserche On Social Network Based On Structural Partial-ordered Attribute Diagram

In recent years, with the prevalence and widely popularity of Web2.0, the researchand application of Social Relation Network has becoming a hot point. Social network datamining is a process to obtain valid, novel, potential information from the massive data andit relates to semi-supervised learning, pattern recognition, database, knowledge acquisition,data visualization and other fields. Researching on the Social Relation Network is helpfulto get the activity patterns in individuals and the development law of social, hasapplication value in reality.Data visualization can help people understand the relationship in data by usinggraphs. The effectiveness of visualization depends on the readability of the result graph.Rapidness and clearness of reflecting the relationship in objects is the basic requirementsof massive social network data mining techniques. However, the traditional social networkvisualization techniques can not clearly reflect the classification hierarchy andrelationships among the various nodes. Therefore, this paper proposes a new method ofstructural partial-ordered attribute diagram, which based on the Formal Concept Analysis(FCA). Its basic idea is to create concept hierarchy structure depends on the binaryrelationship among objects and attributes. It is a mathematical method based on themathematical expression of the philosophical “concept”, human-centered， andconstructing the formal context to discover concepts. At the same time, the method usesvisualization tools to show the overall structure of concept and the association amongconcepts. Thus it can integrate numerous data mining measures to dig out valuableknowledge from massive complex data. Therefore FCA is considered a powerful tool fordata mining and visualization.The Hayes-Roth Data and KRACKHT Data are selected from the social field in thispaper, and be analyzed by the structural partial-ordered attribute diagram. Firstly, theobjects and attributes from the data are defined to construct the formal context, then,structural partial-ordered attribute diagram is produced based on the FCA to analyze thedata, compared with the traditional tool of Pajek in this paper. The results of data mining proved the advantages of partial-ordered attribute diagram and find the knowledge hiddenin the data.