Research On Attribute Reduction And Rules Extraction In Decision Formal Context

With the vigorous development of information technology, especially the constantinnovation and popularization of network technology, the various data has beenproliferating. When face the vast amounts of data, which gives the information processingtechnology in the field of knowledge discovery a new severe challenge, that has becomean issue urgent need to be resolved. In this paper, based on the existing achievements ofresearch and analysis, propose a new partial-ordered attribute diagram to deal with theproblem of attribute reduction.Formal concept analysis, after first raised, has been the rapid development, is usedfor data analysis and knowledge discovery. The attribute reduction is an important topic offormal concept analysis; take the decision formal context as the main subject in this study.Based on the theoretical basis of the formal concept analysis, according to study theconstruction of partial-ordered attribute diagram, explore the solution of attributereduction and rule acquisition.Firstly, according to the theory of formal concept analysis and rough set, there aresuch denotations as the equivalent class, partition, formal concept, formal lattice, anddifferent attributes play different roles in the expression of the objects, investigate thedefinitions about the attribute and the relevant principles of the attribute reduction. Then,apply the exchange procedure of procession of the formal concept of this paper to stratifyoptimization the data set, and formulate the partial-order attribute diagram to describe thecorresponding formal context. Finally, under the premise of fully satisfying the attributeset after reduction can describe the whole object set stilly, put forward the way of theattribute reduction in the field of the decision formal context to extract classification rules.Compared with traditional concept lattice, the partial-order attribute diagrampresented by this article has some odds, for example, smaller amount of calculation,clearer hierarchies, and better visualization. This style is a simple way to explore of theredundant attribute set and decision classify. Though some numerical experiments, dem-onstrate that this diagram can work for the research of the attribute reduction of thedecision formal context.