Knowledge Acquisition And Application Research Based On Formal Concept Analysis

Formal Concept Analysis (FCA) is proposed by Wille in 1982. FCA has a good mathematical property, it not only can describe the relationship between attributes and objects essentially, also can show the relationship of generalization and specialization of concepts intuitively. Therefore, FCA has become a powerful tool for data mining, information retrieval and software engineering.This paper focuses on studying the knowledge expansion of FCA.We discussed the integration of formal concept analysis and rough set, fuzzy set respectively, it theoretically enriched the formal concept analysis theory. On this basis, we explore the knowledge acquisition based on consistent concept lattice and the topic classification method of text based on fuzzy concept lattice. It realized the application value of extension theory.The main content is as follows:(1) Knowledge expansion based on formal concept analysisIn this paper, we combined the theory of formal concept analysis and rough seU fuzzy set respectively, which are describe the uncertain information. We put forward the consistent concept and consistent concept lattice by introducing formal concept analysis in rough set, similarly, we construct the fuzzy concept lattice by combining fuzzy set theory and formal concept analysis. It not only enriches the theory of formal concept analysis, but also provides the theoretical basis for the later application research.(2) Knowledge acquisition and Application based on consistent concept latticeIn this paper, we studied knowledge acquisition in the incomplete information system based on consistent concept lattice, and investigated the application in the incomplete decision table. In the incomplete information system, the lower and upper approximation operators are redefined. For some basic problems in rough set, such as reduction, independent, nuclear, this paper puts forward a new solution. In addition, this paper defines importance factor of attributes and provides a better understanding of decision rule and decision dependency. It is better for us to understand rough set from the perspective of formal concept analysis, and it provides a new way of solving some basic problems in the incomplete information system.(3) Topic classification method of text based on fuzzy concept latticeThis paper presents a Topic classification method of text based on the fuzzy concept lattice. It conceptualizes documents into a more abstract form of concepts, and uses these as the training examples. In order to improve the quality of training data and promote the classification accuracy, we set the concept granularity threshold to get credible concept to eliminate noise, Then the K neighbors is acquired to achieve the classification of texts. The experimental results indicate superior performance. Experimental result shows that the optimal F1 value reached 91.44%. By comparing with the classical K nearest neighbor method and support vector machine method, the result shows that this method is better than KNN and SVM method, the effectiveness of the method is further examined.