Computing Stable Extensions of Argumentation Frameworks using Formal Concept Analysis

Sergei ObiedkovSergei Obiedkov, Barış SertkayaBarış Sertkaya

ISBN: 978-3-031-43618-5
Slides

Sergei Obiedkov, Barış Sertkaya
Computing Stable Extensions of Argumentation Frameworks using Formal Concept Analysis
In Sarah Gaggl, Maria Vanina Martinez, Magdalena Ortiz, eds., Logics in Artificial Intelligence. JELIA 2023, volume 14281 of LNAI, 176–191, September 2023. Springer

Details
Bibtex

KurzfassungAbstract
We propose an approach based on Formal Concept Analysis (FCA) for computing stable extensions of Abstract Argumentation Frameworks (AFs). To this purpose, we represent an AF as a formal context in which stable extensions of the AF are closed sets called concept intents. We make use of algorithms developed in FCA for computing concept intents in order to compute stable extensions of AFs. Experimental results show that, on AFs with a high density of the attack relation, our algorithms perform significantly better than the existing approaches. The algorithms can be modified to compute other types of extensions, in particular, preferred extensions.
Projekt:Project: SECAI, ScaDS.AI
Forschungsgruppe:Research Group: Wissensbasierte SystemeKnowledge-Based Systems

The final publication is available at Springer.

@InProceedings{10.1007/978-3-031-43619-2_13,
author="Obiedkov, Sergei
and Sertkaya, Bar{\i}ş",
editor="Gaggl, Sarah
and Martinez, Maria Vanina
and Ortiz, Magdalena",
title="Computing Stable Extensions of Argumentation Frameworks using Formal Concept Analysis",
booktitle="Logics in Artificial Intelligence",
year="2023",
publisher="Springer Nature Switzerland",
address="Cham",
pages="176--191",
abstract="We propose an approach based on Formal Concept Analysis (FCA) for computing stable extensions of Abstract Argumentation Frameworks (AFs). To this purpose, we represent an AF as a formal context in which stable extensions of the AF are closed sets called concept intents. We make use of algorithms developed in FCA for computing concept intents in order to compute stable extensions of AFs. Experimental results show that, on AFs with a high density of the attack relation, our algorithms perform significantly better than the existing approaches. The algorithms can be modified to compute other types of extensions, in particular, preferred extensions.",
isbn="978-3-031-43619-2"
}