The Second Answer Set Programming Competition [chapter]

Marc Denecker, Joost Vennekens, Stephen Bond, Martin Gebser, Mirosław Truszczyński
2009 Lecture Notes in Computer Science  
This paper reports on the Second Answer Set Programming Competition. The competitions in areas of Satisfiability checking, Pseudo-Boolean constraint solving and Quantified Boolean Formula evaluation have proven to be a strong driving force for a community to develop better performing systems. Following this experience, the Answer Set Programming competition series was set up in 2007, and ran as part of the International Conference on Logic Programming and Nonmonotonic Reasoning (LPNMR). This
more » ... ond competition, held in conjunction with LPNMR 2009, differed from the first one in two important ways. First, while the original competition was restricted to systems designed for the answer set programming language, the sequel was open to systems designed for other modeling languages, as well. Consequently, among the contestants of the second competition were a CLP(FD) team and three model generation systems for (extensions of) classical logic. Second, this latest competition covered not only satisfiability problems but also optimization ones. We present and discuss the set-up and the results of the competition. problem. This idea of model generation as a declarative problem solving paradigm has been pioneered in the area of ASP [1-3]. ASP has three fundamental characteristics: a modeling language based on the syntax of logic programs, the use of the answer set semantics [4] to interpret programs in that language, and a problem-solving methodology in which a program is written so that its answer sets provide solutions. ASP has its origins in Logic Programming (LP) [5, 6] , in particular in the attempts in the 1980s to develop a declarative semantics for logic programs with negation and to turn this logic into a formalism suitable for knowledge representation. Gelfond and Lifschitz sought inspiration in nonmonotonic reasoning [7, 8] and proposed to interpret logic programs as special default theories under the semantics of Reiter [8] . Based on this view, they developed the stable model semantics for logic programs [9] , and extended it later to the answer set semantics for disjunctive logic programs with classical negation [4] , which is the core ASP language today. The area made a major leap in 1997, when the first two systems to compute answer sets of logic programs were developed: dlv [10] and smodels [11] . These systems demonstrated that effective tools for processing answer set programs are possible. Following that, in 1999, Marek and Truszczyński [1] and Niemelä [2] proposed answer set computation as a new declarative problem solving paradigm, and Lifschitz dubbed the area answer set programming [12] . It turned out that a rich class of problems could be modeled elegantly as answer set programs according to this paradigm. That became a strong driving force for the development of fast computational techniques in ASP, and for studies of practical applications where the ASP tools could be used effectively. Experience in areas concerned with checking propositional or Pseudo-Boolean (PB) satisfiability and evaluating Quantified Boolean Formulas (QBF) shows that a programming competition gives an effective incentive to the research community to work on developing better performing systems. ASP sought to emulate that experience. The first preliminary competitions for ASP systems were held in 2002 and 2005 at two Dagstuhl meetings [13] . In 2007, the First Answer Set Programming System Competition [14] was organized as part of LPNMR. That competition consisted of four tracks. In three tracks, solvers were tested on prespecified answer set programs. In the SCore-v and SCore tracks, input consisted of a ground logic program, respectively with and without disjunction; in the SLparse track, the programs used lparse's output language (including aggregates). In the fourth track, called Model, Ground, Solve (MGS), contestants encoded problems in a language of their choice. Ten teams competed. The clasp solver won the SCore and SLparse tracks, and the competition version of dlv won in SCore-v and MGS. In the past two years, some important developments have occurred in ASP. First, as it is clear from the results of the second ASP competition (see Section 5), existing systems have improved considerably, both in available language features and in the speed of the solvers. In addition, new systems have been built, and more teams competed. Clearly, the first ASP competition has had its desired effect! Second, the ASP community has been gradually opening up to other domains. The fields of SAT [15], SAT Modulo Theories (SMT) [16] , CP [17] and, in some way, also Abductive Logic Programming (ALP) [18] are in one key respect very close to ASP. Namely, the ASP declarative problem solving paradigm does not depend on the answer set programming
doi:10.1007/978-3-642-04238-6_75 fatcat:qrm2hfnpyfaxzfricvviuif2xu