Defaults with priorities (Q2642474)

This paper considers rule-based default logics supplemented by a priority relation between the rules. The problem is how best to use this relation to modulate the generation of outputs. As the author recognizes, this is not a new subject. A number of proposals have been made in the literature, many of which are compared in the recent review of \textit{J. P. Delgrande, T. Schaub, H. Tompits} and \textit{K. Wang} [``A classification and survey of preference handling approaches in nonmonotonic reasoning'', Comput. Intell. 20, No. 2, 308--334 (2004)]. The approach adopted in the paper under review is rather different from most of the preceding ones, in that it is directly inspired by work in the neighbouring area of defeasible inheritance, in which the author has himself played an important role. The constructions developed in the paper are `tested' by running them on a range of examples, with outcomes compared with those obtained under alternative accounts. A particularly interesting example, which appears to be new to the default logic literature, is that of the `ruffed finches'. In this example, prioritization between rules is given by their relative specificity, and intuitively we appear to be ready to suspend the application of a default rule in favour of another even when the defaults do not conflict. As the author points out, this has for some time been recognized in the literature of another neighbouring field, the logic of defeasible argumentation, where it is sometimes called `undercutting' (see e.g. [\textit{J. Pollock}, Cognitive carpentry. Cambridge MA: MIT Press (1995)]). Although the phenomenon is highlighted by the author, it is not taken account of in any of his constructions, but rather included in a list of open problem areas given at the end of the paper.