The Research Of Concept Lattice Pruning Method And Its Application In Web Mining

Formal concept analysis is an effective tool for the conceptual knowledge processing. It developed widely and used in many fields, such as the knowledge discovery, software engineering, information retrieval, database management, web mining and so on. Along with the rapid development network technologies and the increase of the size of the database, not only valuable information but also massive redundant information may exist in these databases. Thus the study of how to extract useful knowledge from massive information effectively has great practical value.The concept lattice construct for large databases has too much nodes and become large scale, and it becomes difficult to analyze so that lose its value. Therefore, it's need to reduce the size of concept lattices, making it more clear reflection of the key information in the database. This paper studied the basic theory of concept lattice, some kinds of the concept lattice construction algorithm, and study deeply on the purring theory of concept lattice and its application in web mining. First, we introduced two pruning strategies in this paper:support pruning and concept stability pruning. In the concept lattice, the support pruning method is mining frequent itemsets to establish iceberg concept lattices, but it may be hidden those do not frequent concept with practical significance. However, the concept stability purring method is based on the correlation of the concept in the concept lattice. Then, it proposed a algorithm to calculate the stability of concepts, and analyzes the concept of stability theory. At last, the formal concepts of stability theory combine with application of web mining, and use the concept lattice to analysis the folksonomy, which can reflect the acquaintance of this community, it's beneficial to organize and use the resources in each community. To contrast the support purring and stability purring on the concept lattice, we did some experiments on folksonomy and it indicate that stability measurement can remove redundant information and find users' interesting.