Association Rules Detecting Based On Attribute Topology

Formal concept analysis theory is an important method of knowledge discovery,because the concept is defined strictly and concept lattice describes the relationship between concepts clearly.AT(Attribute Topology)theory emphasizes the binary relationship between attributes,highlighting the visualization,ordering and parallelization of concept computing.Formal concept as a form of association rules,pay attention to the important link between the property objects,this requirement is sometimes too strict.Therefore,it is necessary to introduce other types of association rules,rich the AT theory.The association rules among attributes can be divided into three types: frequent association rules based on frequence,partial order association rules based on the localized importance,granular association rules based on the level of granular,causal association rule based on necessity,formal concept analysis.Based on this,the following things has been finished.Firstly,frequent association rules mining is proposed in AT.AT and FP-tree can be transformed bidirectionally,which can be proved by making use of the binary relationship between attributes,and a transformation algorithm is proposed.The mining of frequent association rules in AT is realized with the classical FP-tree model.Based on AT,a visual algorithm for mining rules has been proposed furtherly.Secondly,partial order association rules mining is proposed in AT.AT and Attribute Partial Order can be transformed by means of the equivalent deformation between binary relations.Based on this,the binary transformation algorithm has been proposed.Finally,granularity association rules mining is proposed in AT.The definition of Topology Granular has been proposed,and an analysis method of AT's granular structure is proposed.At the multi-granularity level,analysisation and extraction of association rules in and between Topology Granular has been achieved.These three aspects,complemented with each other,construct the preliminary association discovery algorithm system in AT.