Untitled

Abstract

In this paper we focus on the general problem of identifying and resolving conflicts such as inconsistency and incompleteness in scientific discovery. The underlying idea is that conflicts in empirical knowledge can lead to new discoveries, provided that they are clearly identified. Our observations are based on the behavior and the results of an incremental discovery program, BR-3. The system models the discoveries of a series of quantum laws by physicists in this century. BR-3's discoveries are directed by a small set of consistency and completeness constraints. We evaluate the system in terms of its knowledge representation, discovery operators, and their behavior, and we describe its relation to other work in scientific discovery.