Fusing Approximate Knowledge from Distributed Sources

In this paper we investigate a technique for fusing approximate knowledge obtained from distributed, heterogeneous information sources. We use a generalization of rough sets and relations [14], which depends on allowing arbitrary similarity relations. The starting point of this research is [2], where a framework for knowledge fusion in multi-agent systems is introduced. Agent’s individual perceptual capabilities are represented by similarity relations, further aggregated to express joint capabilities of teams. This aggregation, allowing a shift from individual to social level, has been formalized by means of dynamic logic. The approach of [2] uses the full propositional dynamic logic, not guaranteeing the tractability of reasoning. Therefore the results of [11, 12, 13] are adapted to provide a technical engine for tractable approximate database querying restricted to a Horn fragment of serial PDL. We also show that the obtained formalism is quite powerful in applications. 1 Similarities and Approximate Reasoning In this paper we investigate a technique for fusing approximate knowledge obtained from distributed information sources. This issue is substantial in modeling multiagent systems, where a group of loosely coupled heterogeneous agents cooperate in B. Dunin-Kȩplicz Institute of Informatics, Warsaw University, Poland and ICS, Polish Academy of Sciences, Poland e-mail: [email protected] L.A. Nguyen Institute of Informatics, Warsaw University, Poland e-mail: [email protected] A. Szałas Institute of Informatics, Warsaw University, Poland and IDA, Linköping University, Sweden e-mail: [email protected] ∗ Supported by the MNiSW grants N N206 399134 and N N206 399334.