A metadata architecture for case-based reasoning

Software design architectures created to meet generic requirements with specific requirements expressed in metadata are referred to as metadata architectures. In this paper we discuss a metadata architecture for a case-based reasoner using a reflective mechanism for implementation. The important aspect of this paper is that the it defines a reasoning system that is independent of the specific application domain, but which can be instantiated to create a reasoner for some given domain via specification of metadata appropriate for that domain. Thus the system can be used for many different types of domains. To date, it has been used to create case-based reasoners for (1) detecting possible intrusion packets in a network using past packet cases and their possible threats, (2) matching of a face in a facial case base using Eigenvalue features, and (3) solving network failure problems based on previous failure cases and their solutions.