Some observations on modelling case based reasoning with formal argument models
Trevor J. M. Bench-Capon · 1999
In this paper I shall explore the modelling of case based reasoning using a formal model of argument, taking the approach of Prakken and Sartor as my starting point. I first consider their method of representing cases, and describe how — if we restrict ourselves to independent boolean factors — we can fruitfully model the domain as a partial order on rules. I then consider the issues relating to quantifiable factors, as used in HYPO, and factor hierarchies, as used in CATO. The former presents some difficulties for modelling as a partial order, and, coupled with the latter, forces us to recognise two different kinds of reasoning used in concept application which have different implications for representing the domain. I then present some conclusions arising from the discussion.