Article ID | Journal | Published Year | Pages | File Type |
---|---|---|---|---|
9952065 | Artificial Intelligence | 2018 | 4 Pages |
Abstract
ASPIC+ is a general framework for structured argumentation that allows for a considerable degree of freedom as to the knowledge representation choices made in specifying the premises and rules from which arguments are constructed, and the preferences used to determine which attacks amongst constructed arguments succeed as defeats. On the other hand, the framework provides guidelines for making these choices, such that adherence to these guidelines suffices to guarantee satisfaction of the consistency and closure postulates. In particular, guidelines are given for selecting ways in which to compare the sets of defeasible constituents of arguments in order that the defined preference relations over arguments guarantee satisfaction of the rationality postulates. However, it has recently been noted that the way in which these set comparisons are made, and the guidelines for choosing these comparisons, admit counter-examples to these postulates. This research note proposes simple revisions to the ASPIC+ framework as defined in [3]. In this way, the counter-examples are avoided and rationality is preserved.
Keywords
Related Topics
Physical Sciences and Engineering
Computer Science
Artificial Intelligence
Authors
Sanjay Modgil, Henry Prakken,