Explaining Answer Sets using Argumentation Theory

Talk by Claudia Schulz

Location: APB 0005
Start: 28. November 2014 at 3:00 pm
End: 28. November 2014 at 4:00 pm
Research group: Computational Logic
iCal

One of the prominent techniques for solving knowledge representation and reasoning problems is answer set programming (ASP). A problem is encoded as a set of inference rules expressing everything known about this problem, and the problem's solutions, the answer sets, are the sets of all non-conflicting literals deducible from these rules. Answer sets can be efficiently computed using answer set solvers; however they do not provide any explanation as to why a literal is or is not part of this answer set. Having an explanation of literals in a solution is particularly desirable when ASP is used as a reasoning tool in applications such as medical decision making, where the solutions are used by non-ASP-experts like doctors.

In this talk I will present ABA-Based Answer Set Justifications, which based on Argumentation Theory explain why a literal is or is not contained in an answer set. Reasoning in Argumentation Theory involves forming arguments from the given knowledge and evaluating the conflicts between them. Since this type of reasoning is easily understandable for humans yet processable by computers, Argumentation Theory is a suitable technique for providing explanations of answer sets for humans. ABA-Based Answer Set Justifications are based on the correspondence between answer sets and stable extensions in Argumentation Theory and provide an explanation in terms of an admissible fragment of the stable extension corresponding to the answer set in question.

More info at: http://www.doc.ic.ac.uk/~cis11/