A framework for the management of past experiences with time-extended situations

Abstract

In the context of knowledge management, we focus on the representation and the retrieval of past experiences called cases within the Case-Based Reasoning (CBR) paradigm. CBR is a problem-solving method based on the reuse of past experiences that represent pieces of knowledge. The first step of this kind of reasoning is the retrieval from the memory of relevant cases to solve a new problem by reuse and adaptation. In this paper, we propose a framework to be used in the design of CBR systems that need a retrieval of cases with timeextended situations. Time-extended situation assessment is required in many different applications such as automatic control, medical problem solving, process supervision and forecasting. Inside our framework for past experiences management, we propose a representation model that mainly handles situations described by a set of sequences of events and/or sampled data. We also propose a complex retrieval strategy that integrates the retrieval of different types of knowledge: abstract, concrete and potential cases. Our framework, implemented in an object-oriented way, can be applied to a general class of problems and mainly supports the knowledge discovery and explanation.