Introducing CoQuiAAS

CoQuiAAS is an open-source software developed to answer a current need in the Artificial Intelligence community : the ability to solve efficiently inference tasks about argumentation frameworks. Our software provides the sufficient reasoning tools to work with the usual acceptability semantics (Complete, Preferred, Stable and Grounded) and the usual inference tasks (compute one extension, enumerate all the extensions, credulous acceptance and skeptical acceptance).
CoQuiAAS respects the requirements of the First International Competition on Computational Models of Argumentation and can use the different input formats of the competition.
CoQuiAAS has been developed jointly by three members of the CRIL (Centre de Recherche en Informatique de Lens) and LIPADE (Laboratoire d'Informatique Paris Descartes).
CoQuiAAS is supported the project AMANDE, funded by the French National Research Agency (ANR) as ANR-13-BS02-004 and the Fondation de Recherche pour l'Aéronautique et l'Espace (FRAE).

News

Competition Results: The Award Ceremony of ICCMA 2015 took place during the workshop TAFA'15 in Buenos Aires. CoQuiAAS has received the award of "First Place", being ranked first among 18 solvers. Details are available on ICCMA'15 website.

Basics of Abstract Argumentation

Abstract Argumentation Frameworks (AFs) [Dun95] are directed graphs where the nodes are so-called arguments, and the edges represent some attacks.
Several acceptability semantics allow to decide which sets of arguments, called extensions, are the acceptable outcome of the AF. For these semantics, several inference problems can be studied:

  • Computing one extension of a given AF (SE)
  • Enumerating all the extensions of a given AF (EE)
  • Decide if a given argument is credulously accepted (ie. belongs to at least one extension) by a given AF (DC)
  • Decide if a given argument is skeptically accepted (ie. belongs to each extension) by a given AF (DS)
These are the tasks which are currently performed by CoQuiAAS, for the complete (CO), preferred (PR), stable (ST) and grounded (GR) semantics.

How to use it

CoQuiAAS complies with the command-line requirements of the First International Competition on Computational Models of Argumentation:
The possible input formats are described here:

  • ASPARTIX format (apx): each argument is defined by arg(‹name of the argument›) and each attack is defined by att(‹name of the attacking argument›,‹name of the attacked argument›), one by line. Each line finishes with a dot.
    arg(a).
    arg(b).
    arg(c).
    att(a,b).
    att(b,a).
    att(b,c).
  • Trivial Graph Format (tgf): each argument is defined by its name, one by line. These definitions are followed by a sharp symbol. Then the attacks are defined by the name of the attacking argument, a blank space and the name of the attacked argument. Each line is a single attack.
    a
    b
    c
    #
    a b
    b a
    b c
  • Dimacs format (cnf):
    p af 3 3
    1 -2 0.
    2 -1 0
    2 -3 0
The command-line arguments are:
  • -p XX-YY where
    • XX in {EE,SE,DC,DS}
    • YY in {CO,PR,ST,GR}
  • -fo format where format in {apx,tgf,cnf}
  • -f ‹file path›
  • -a param where param is an argument name (useful for credulous and skeptical acceptance)
CoQuiAAS can be used as a library in another software. Help to use it can be found in its documentation.
Currently, CoQuiAAS incorporates the software coMSSExtractor to perform the constraint-based process; it uses the approach described in [GLM14].

Experimental results

To test our software, we used the benchmarks provided by the organizers of the competition: here and there. The table below presents the average times (rounded to 0.01 second) by family for CoQuiAAS to enumerate the extensions of the AFs, for each family of benchmark provided. The symbol - means that the whole family reaches the timeout (900 seconds), in the other case, the whole family is solved.

Average time by family to enumerate the extensions for each semantics
Family#Inst.GrStPrCo
rdm2025‹0.010.01‹0.01‹0.01
rdm5024‹0.01‹0.01‹0.01‹0.01
rdm20024‹0.010.55.321.57
rdm100060.25---
real207.257.518.556.88
XXX2004‹0.010.080.040.03
XXX300640.0212.5334.821
XXX400220.010.120.130.08

Downloads

ICCMA'17 version (fixed):

  • CoQuiAAS (ICCMA'17) ; in order to use PR, SST, STG and ID semantics, you must download the LBX solver and provide its path to CoQuiAAS through the "-lbx" parameter.

Compiled version of CoQuiAAS:

Sources files of CoQuiAAS (GNU Automake):

Related Publications

  • International Conference: Jean-Marie Lagniez, Emmanuel Lonca, Jean-Guy Mailly, « CoQuiAAS: A Constraint-based Quick Abstract Argumentation Solver » , Twenty-seventh IEEE International Conference on Tools with Artificial Intelligence (ICTAI'15) - cite
  • National Audience Conference: Jean-Marie Lagniez, Emmanuel Lonca, Jean-Guy Mailly, « CoQuiAAS : Applications de la programmation par contraintes à l'argumentation abstraite », Onzièmes Journées Francophones de la Programmation par Contraintes (JPFC'15)
  • System Description: Jean-Marie Lagniez, Emmanuel Lonca, Jean-Guy Mailly, « CoQuiAAS: Application of Constraint Programming for Argumentation », First International Competition on Computational Models of Argumentation (ICCMA'15)

Contact

CRIL-CNRS UMR 8188
Université d'Artois
Faculté des Sciences Jean Perrin
Rue Jean Souvraz SP 18
62307 LENS Cedex (France)

 

UFR de Mathématiques et Informatique
LIPADE - Équipe IAD
Université Paris Descartes
45 rue des Saints Pères
75006 PARIS

Bibliography

[Dun95]   Phan Minh Dung. On the acceptability of arguments and its fundamental role in nonmonotonic reasoning, logic programming, and n-person games. Artificial Intelligence, 77(2):321-357, 1995.
[GLM14]   Éric Grégoire, Jean-Marie Lagniez, Bertrand Mazure. An Experimentally Efficient Method for (MSS, CoMSS) Partitioning. AAAI-14, 2666-2673, 2014.