The Construction Of Concept Based On The Maximal Closed Itemset And Its Application

With the development of Internet technology and the surge of the data in many areas, the data mining technology and the knowledge discovery technology play more and more important role in the practical application. At the same time, they bring new challenges and opportunities for research in these areas. The theory of formal concept analysis provides new theoretical basis for the data mining technology and the knowledge discovery technology. In this paper, the construction method and its application of formal concept is studied.First, an constructing algorithm of binary concept which is looking for the maximal closed 1- matrix in the two-dimensional cross table is designed,based on the theory of maximal closed itemset and the peer relationship between binary concept and the maximal closed itemset. That is looking for the maximal closed matrix of the relation of object and attribute is "1",in the two-dimensional cross table. Through the example analysis, the correctness and feasibility of the algorithm is proved.Secondly, if the constructing algorithm of binary concept based on the maximal closed 1- matrix is extended to triadic concept construction, a lot of cuboids can be finded which are consist of "1" and "0" in the three-dimensional cross table. And the maximal closed cuboid which is consist of "1" has one-to-one correspondence relationship with triadic concept. So the triadic concept construction algorithm is designed based on the extension matrix, and the algorithm is optimized. Through the example analysis, the correctness and feasibility of the algorithm is proved.Finally, the application of formal concept analysis in the folksonomy system is studied. Through the research of user behavior and user preference in the folksonomy system, the user group division is realized from resource, user, and label three perspectives. And user preference for resource is studied from individual resource and resource preference set two perspectives. The results of the analysis show that the formal concept analysis method has a better sense of hierarchy and visualization and so on.