Reasoning About Priorities in Default Logic

Abstract

In this paper we argue that for realistic applications involving default reasoning it is necessary to reason about the priorities of defaults. Existing approaches require the knowledge engineer to explicitly state all relevant priorities which are then handled in an extra-logical manner, or they are restricted to priorities ba-sed on specificity, neglecting other relevant criteria. We present an approach where priority information can be represented within the logical language. Our approach is based on PDL, a prioritized exten-sion of Reiter’s Default Logic recently proposed by the same author. In PDL the generation of extensions is controlled by an ordering of the defaults. This pro-perty is used here in the following way: we first build Reiter extensions of a given default theory. These ex-tensions contain explicit information about the prio-rities of defaults. We then eliminate every extension E that cannot be reconstructed as a PDL extension based on a default ordering that is compatible with the priority information in E. An example from legal reasoning illustrates the power of our approach. 1.

Cite

Text

Brewka. "Reasoning About Priorities in Default Logic." AAAI Conference on Artificial Intelligence, 1994.

Markdown

[Brewka. "Reasoning About Priorities in Default Logic." AAAI Conference on Artificial Intelligence, 1994.](https://mlanthology.org/aaai/1994/brewka1994aaai-reasoning/)

BibTeX

@inproceedings{brewka1994aaai-reasoning,
  title     = {{Reasoning About Priorities in Default Logic}},
  author    = {Brewka, Gerhard},
  booktitle = {AAAI Conference on Artificial Intelligence},
  year      = {1994},
  pages     = {940-945},
  url       = {https://mlanthology.org/aaai/1994/brewka1994aaai-reasoning/}
}