We focus on realistic, complex domains where the concepts, terms and predicates used by domain rules or by rule-based models are not well-defined. Often, in such inherently ill-defined domains the rules do not encompass all the situations they are asked or assumed to cover, admit tacit exceptions, or can be contradicted and annulled by other rules. Interpretation is therefore required of the terms and predicates used. The law is a prototypical example of such an area, where terms used in legal statutes are not completely defined by legal regulations. The use of case-based reasoning (CBR) to complement and supplement other types of reasoning involves many computational questions of system architecture and control. The key focus of this work is how and when to interleave CBR with other modes of reasoning in the context of applying a rule or model to a new set of facts in light of a corpus of cases of past application. The goal is to generate an explanation or argument as to how the new fact situation might be interpreted. In particular, we report on a system called CABARET (CAse-BAsed REasoning Tool), a hybrid architecture we have built to study and experiment with these issues.

The next project was Joel Moses's program SIN. See Moses, Symbolic Integration: The Stormy Decade, 14 Comm.

This paper presents a hybrid case-based reasoning (CBR) and information retrieval (IR) system, called SPIRE, that both retrieves documents from a full-text document corpus and from within individual documents, and locates passages likely to contain information about important problem-solving features of cases. SPIRE uses two case-bases, one containing past precedents, and one containing excerpts from past case texts. Both are used by SPIRE to automatically generate queries, which are then run by the INQUERY full-text retrieval engine on a large text collection in the case of document retrieval and on individual text documents for passage retrieval.

We discuss key ingredients of case-based reasoning, in general, and 3. A technique, the "claim lattice", for organizing the the correspondence of these to elements of HYPO.

Most recent work on reasoning with cases in law has taken the style of reasoning used in the CATO system as its model, and uses the notion of factors, as found in that system. Fundamental to CATO, a successor the HYPO system, were factors, which are closely related to HYPO-style dimensions. In this paper, we will argue that the simplification involved in using factors, while it has proved pragmatically useful both for clarifying understanding of certain aspects of reasoning with cases and for implementation, causes problems with domain analysis and precludes certain kinds of argument that we would wish to model. We therefore believe that the time is now ripe to go back to the original notion of dimensions, while retaining the insights that have been gained from working with the simpler notion of factors. The paper uses two case studies to argue that this is so.

Scripts, Plans, Goals, and understanding.

For In this paper we discuss a general approach to detecting instance, a single Supreme Court case created the conceptual change developed in our on-going work on "automobile exception" to the Fourth Amendment's warrant concept drift [Rissland et al., 1994]. We illustrate our requirement for a constitutionally acceptable search; this approach on an actual, still evolving, legal example, the case forever changed the meaning of our Fourth "good faith" concept in the law of personal bankruptcy. In Amendment, which is still evolving today [Rissland, 1989; our approach, we detect that a concept is changing by Rissland & Collins, 19861.

We use the idea of a "dimension," how they are generated.

INTRODUCTION There is mounting evidence that human experts rely heavily on memory of past cases when solving problems in domains such as law, mathematics, design, and strategic planning. Thus, it seems natural to exploit this idea in constructing Al systems. This is the focus of systems using case-based reasoning; it constitutes a fifth major paradigm of machine learning research. A related approach is that of reasoning by analogy. In case-based reasoning ("CBR"), one uses memory of relevant "past" cases to interpret or to solve a new problem case. Rather than creating a solution from scratch, a reasoner using case-based reasoning recalls cases similar to its current problem situation and solves or interprets a problem by reasoning with past solutions and interpretations. A reasoner using case-based reasoning can derive shortcuts and anticipate problems in new situations that might arise by having previously spotted and dealt with them. This can lead to improvement in the quality and efficiency of the reasoning. Case-based reasoning as a learning paradigm has several advantages. First, there are several performance enhancements it provides for its associated performance element: shortcuts in reasoning, the capability of avoiding past errors; the capability of anticipating and therefore avoiding other previously made mist akes, the capability of focusing in on the most important parts of a problem first. Second, learning can be fairly uncomplicated.

In this paper, we consider a distinctly bottom-up which generates arguments by performing a heuristic approach.