A FAST ALGORITHM FOR GENERATING CONCEPTS USING AN ATTRIBUTE TABLE

Asanunsupervisedlearningtechniqueforconceptualclustering,FormalConcept Analysishasbeenwidelyusedinmanyareasincludingmachinelearninganddata mining. This paper organizes the search space for concepts as a prefix tree, and employs a new data structure called Attribute Table to map and prune the tree. Thus, the procedures of conceptual clustering can be carried out only in a few valid subspaces. A fast algorithm for generating concepts is proposed in this paper. Experimental evidence shows that our algorithm performs very well for generating concepts on both dense and sparse contexts.