Talks and Poster Presentations (with Proceedings-Entry):
A. Haret, J. P. Wallner:
"Manipulation of Semantic Aggregation Procedures for Propositional Knowledge Bases and Argumentation Frameworks";
Talk: International Workshop on Non-Monotonic Reasoning (NMR),
- 2018-10-29; in: "Proceedings of the 17th International Workshop on Non-monotonic Reasoning",
E. Fermé, S. Villata (ed.);
Paper ID 146,
We study the potential for manipulation of semantic aggregation procedures for argumentation frameworks (AFs) and propositional knowledge bases. The basis for our study is a framework that investigates questions of manipulability and strategy-proofness for logic-based merging, which provides the foundation to several operators for aggregating AFs. We enrich a key component of this framework---the satisfaction indices which formalize when manipulation by an agent pays off---with two main approaches to AF reasoning: skeptical and credulous acceptance of arguments. In propositional knowledge bases, skeptical consequences are propositions true in all models, while credulous consequences are propositions true in at least one model. We find that, even in restricted cases, most aggregation procedures are vulnerable to manipulation by an agent for both (i) propositional knowledge bases and (ii) AFs under the stable, preferred, and grounded semantics. However, aggregation is not manipulable (i.e., is strategy-proof) under one well-known operator when the bases, or AFs, are semantically complete (i.e., have exactly one semantical model). Finally, we provide complexity results for computing our novel indices.
Project Head Johannes Peter Wallner:
Project Head Stefan Woltran:
Created from the Publication Database of the Vienna University of Technology.