Affiliations: Institute of Computer Science, University of Silesia, Będzińska 39, 41-200 Sosnowiec, Poland. malgorzata.przybyla-kasperek@us.edu.pl; alicja.wakulicz-deja@us.edu.pl
Note: [] Address for correspondence: University of Silesia, Institute of Computer Science, Będzińska 39, 41-200 Sosnowiec, Poland
Abstract: The paper includes a discussion of issues related to the process of global decision-making on the basis of information stored in several local knowledge bases. The local knowledge bases contain information on the same subject, but are defined on different sets of conditional attributes that are not necessarily disjoint. A decision-making system, which uses a number of knowledge bases, makes global decisions on the basis of a set of conditional attributes specified for all of the local knowledge bases used. The paper contains a description of a multi-agent decision-making system with a hierarchical structure. Additionally, it briefly overviews methods of inference that enable global decision-making in this system and that were proposed in our earlier works. The paper also describes the application of the conditional attributes reduction technique to local knowledge bases. Our main aim was to investigate the effect of attribute reduction on the efficiency of inference in such a system. For a measure of the efficiency of inference, we mean mainly an error rate of classification, for which a definition is given later in this paper. Therefore, our goal was to reduce the error rate of classification.
Keywords: decision-making system, global decision, rough set, reduction of set of conditional attributes
