Knowledge discovery objects and queries in Distributed Knowledge Systems

Abstract

The development of many knowledge discovery methods (see [14], [7], [16]) provided us with good foundations to build a kd-Query Answering System (kdQAS) for Distributed Knowledge Systems (DKS). By DKS we mean a number of autonomous processing elements (called knowledge systems) that are interconnected by a computer network and that cooperate in their assigned tasks. A knowledge-system we see as a relational database coupled with a discovery layer which is simplified in this paper to a set of rules.

Queries handled by kdQAS are more general than SQL. Also, the queried objects are far more complex than tuples in a relational database. To distinguish them from objects and queries in DBMS, we introduce kd-objects and kd-queries respectively. In general, by kd — object we mean any set of tuples and rules. By kd — query we mean a predicate which queries kd-object in DKS and returns another kd-object for an answer. Our kd-objects may not exist a priori, thus querying them at one site of DKS may require generation, at run time, of new kd-objects either at the same site or at other sites of DKS. So, querying has to major roles: generation of new kd-objects and retrieval of the ones which were generated before.

In relational databases, the result of a query is a relation that can be queried further. This is typically referred to as a closure principle, and it should be one of the most important design principles for kdQAS. Our kd-queries satisfy such a closure principle.