Answer Set Solving with Generalized Learned Constraints

Conflict learning plays a key role in modern Boolean constraint solving. Advanced in satisfiability testing, it has meanwhile become a base technology in many neighboring fields, among them answer set programming (ASP). However, learned constraints are only valid for a currently solved problem instance and do not carry over to similar instances. We address this issue in ASP and introduce a framework featuring an integrated feedback loop that allows for reusing conflict constraints. The idea is to extract (propositional) conflict constraints, generalize and validate them, and reuse them as integrity constraints. Although we explore our approach in the context of dynamic applications based on transition systems, it is driven by the ultimate objective of overcoming the issue that learned knowledge is bound to specific problem instances. We implemented this workflow in two systems, namely, a variant of the ASP solver clasp that extracts integrity constraints along with a downstream system for generalizing and validating them.

[1]  Barry O'Sullivan.  Automated Modelling and Solving in Constraint Programming , 2010, AAAI.

[2]  Mario Alviano,et al.  Advances in WASP , 2015, LPNMR.

[3]  Yuliya Lierler,et al.  Answer Set Programming Based on Propositional Satisfiability , 2006, Journal of Automated Reasoning.

[4]  Antonius Weinzierl,et al.  OMiGA : An Open Minded Grounding On-The-Fly Answer Set Solver , 2012, JELIA.

[5]  Maurice Bruynooghe,et al.  Detection and exploitation of functional dependencies for model generation , 2013, Theory and Practice of Logic Programming.

[6]  Joao Marques-Silva,et al.  GRASP: A Search Algorithm for Propositional Satisfiability , 1999, IEEE Trans. Computers.

[7]  Krysia Broda,et al.  Inductive Learning of Answer Set Programs , 2014, JELIA.

[8]  Michael Gelfond,et al.  Classical negation in logic programs and disjunctive databases , 1991, New Generation Computing.

[9]  Mary Sheeran,et al.  Checking Safety Properties Using Induction and a SAT-Solver , 2000, FMCAD.

[10]  Martin Gebser,et al.  plasp: A Prototype for PDDL-Based Planning in ASP , 2011, LPNMR.

[11]  Vladimir Lifschitz,et al.  Answer set programming and plan generation , 2002, Artif. Intell..

[12]  Simon Colton,et al.  Automatic Generation of Implied Constraints , 2006, ECAI.

[13]  Sebastian Haufe,et al.  Automated verification of state sequence invariants in general game playing , 2012, Artif. Intell..

[14]  Jussi Rintanen,et al.  An Iterative Algorithm for Synthesizing Invariants , 2000, AAAI/IAAI.

[15]  Ramón P. Otero,et al.  Induction of Stable Models , 2001, ILP.

[16]  Toby Walsh,et al.  Handbook of Satisfiability: Volume 185 Frontiers in Artificial Intelligence and Applications , 2009 .

[17]  Malte Helmert,et al.  Concise finite-domain representations for PDDL planning tasks , 2009, Artif. Intell..

[18]  Miroslaw Truszczynski,et al.  Answer set programming at a glance , 2011, Commun. ACM.

[19]  Sharad Malik,et al.  Boolean Satisfiability Solvers and Their Applications in Model Checking , 2015, Proceedings of the IEEE.

[20]  Roman Barták,et al.  Constraint Processing , 2009, Encyclopedia of Artificial Intelligence.

[21]  Martin Gebser,et al.  Conflict-driven answer set solving: From theory to practice , 2012, Artif. Intell..

[22]  Fangzhen Lin.  Discovering State Invariants , 2004, KR.