Deciding Acceptance in Incomplete Argumentation Frameworks

Authors: Andreas Niskanen, Daniel Neugebauer, Matti Järvisalo, Jörg Rothe2942-2949

AAAI 2020 | Conference PDF | Archive PDF | Plain Text | LLM Run Details

Reproducibility Variable Result LLM Response
Research Type Experimental We continue by an overview of results from an empirical evaluation of the scalability of the approaches to acceptance problems in incomplete AFs presented in this paper. Our implementation uses Glucose 4.1 (Audemard and Simon 2018) as the underlying SAT solver, and is available in open source via https://bitbucket.org/andreasniskanen/ taeydennae. The experiments were run on Intel Xeon E5-2680 v4 2.4-GHz, 256-GB machines with Cent OS 7. We set a per-instance time limit of 900 seconds and a per-instance memory limit of 64 GB.
Researcher Affiliation Academia 1Helsinki Insitute for Information Technology HIIT, Department of Computer Science, University of Helsinki, Finland 2Institut f ur Informatik, Heinrich-Heine-Universit at D usseldorf, Germany
Pseudocode Yes Algorithm 1 CEGAR for possible existence and skeptical acceptance for s {AD =/0, ST}. Input: IAF = A ,A ?,R,R? , a A . 1: ϕ ABSTRACTION(IAF,a) 2: while true do 3: (sat,τ) SAT(ϕ) 4: if sat = false then return reject 5: AF EXTRACT(τ) 6: (sat,τ) SAT(CHECK(IAF,AF ,a)) 7: if sat = false then return accept 8: ϕ ϕ REFINE(IAF,AF ) 9: end while
Open Source Code Yes Our implementation uses Glucose 4.1 (Audemard and Simon 2018) as the underlying SAT solver, and is available in open source via https://bitbucket.org/andreasniskanen/ taeydennae.
Open Datasets Yes We generated incomplete AFs based on the ICCMA 2017 (Gaggl et al. 2016) benchmarks as follows. ... We used the ICCMA 2017 benchmark set A for problems on the second level and the set B for problems on the first level, in-line with the complexity of the acceptance problems for which these sets were used in ICCMA 2017.
Dataset Splits No The paper refers to using ICCMA 2017 benchmarks but does not provide specific details on training, validation, or test dataset splits, such as percentages or sample counts.
Hardware Specification Yes The experiments were run on Intel Xeon E5-2680 v4 2.4-GHz, 256-GB machines with Cent OS 7.
Software Dependencies Yes Our implementation uses Glucose 4.1 (Audemard and Simon 2018) as the underlying SAT solver
Experiment Setup Yes We set a per-instance time limit of 900 seconds and a per-instance memory limit of 64 GB. We generated incomplete AFs based on the ICCMA 2017 (Gaggl et al. 2016) benchmarks as follows. For each AF, we select a query argument uniformly at random from the set of arguments. Now, for each probability p {0.05,0.1,0.15,0.2}, we generated three incomplete AFs: one where each argument (except for the query) is uncertain with probability p, one where each attack is uncertain with probability p, and one where each argument and attack is uncertain with probability p.