Constraint Handling Rules (CHR) is a high-level programming language based on multiheaded multiset rewrite rules. Originally designed for writing user-defined constraint solvers, it is now recognized as an elegant general purpose language. CHR-related research has surged during the decade following the previous survey by Frühwirth (1998). Covering more than 180 publications, this new survey provides an overview of recent results in a wide range of research areas, from semantics and analysis to systems, extensions and applications.

Abstract. As a novel, grand AI challenge, General Game Playing is concerned with the development of systems that understand the rules of unknown games and play these games well without human intervention. In this paper, we show how Answer Set Programming can assist a general game player with the special class of single-player games. To this end, we present a translation from the general Game Description Language (GDL) into answer set programs (ASP). Correctness of this mapping is established by proving that the stable models of the resulting ASP coincide with the possible developments of the original GDL game. We report on experiments with single-player games from past AAAI General Game Playing Competitions which substantiate the claim that Answer Set Programming can provide valuable support to general game playing systems for this type of games. 1

A general game player is a system that understands the rules of an unknown game and learns to play this game well without human intervention. To succeed in this endeavor, systems need to be able to extract and prove game-specific knowledge from the mere game rules. We present a practical approach to this challenge with the help of Answer Set Programming. The key idea is to reduce the automated theorem proving task to a simple proof of an induction step and its base case. We prove correctness of this method and report on experiments with an offthe-shelf Answer Set Programming system in combination with a successful general game player. 1

Many programs for playing games rely on some type of state space search to choose their actions. It is well known that there is a correlation between the depth to which these programs search and the strength of their play. Unfortunately when playing games in the general game-playing competition, manipulating states requires expensive logical inference and as a result, programs that play these games cannot examine many states before they are required to act. Typically, these programs make use of a Prolog package to perform the required logical inference. These packages are not designed specifically for general game playing. As a result, there is great potential for improving the strength of play by designing a specialized package to perform the state manipulation tasks. In this paper we introduce gdlcc, a program that takes a general game description and creates a game specific C++ library for performing state manipulating tasks. Experimental results show that a program using this library, as opposed to a Prolog package, can examine between 60 % and 1760 % more states per action. 1

Abstract. Research in General Game Playing aims at building systems that learn to play unknown games without human intervention. We contribute to this endeavour by generalising the established technique of decomposition from AI Planning to multi-player games. To this end, we present a method for the automatic decomposition of previously unknown games into independent subgames, and we show how a general game player can exploit a successful decomposition for game tree search. 1

Abstract. Unlike traditional game playing, General Game Playing is concerned with agents capable of playing classes of games. Given the rules of an unknown game, the agent is supposed to play well without human intervention. For this purpose, agent systems that use deterministic game tree search need to automatically construct a state value function to guide search. Successful systems of this type use evaluation functions derived solely from the game rules, thus neglecting further improvements by experience. In addition, these functions are fixed in their form and do not necessarily capture the game’s real state value function. In this work we present an approach for obtaining evaluation functions on the basis of neural networks that overcomes the aforementioned problems. A network initialization extracted from the game rules ensures reasonable behavior without the need for prior training. Later training, however, can lead to significant improvements in evaluation quality, as our results indicate. 1

Abstract not found

The Game Description Language is a high-level, rule-based formalisms for communicating the rules of arbitrary games to general game-playing systems, whose challenging task is to learn to play previously unknown games without human intervention. Originally designed for deterministic games with complete information about the game state, the language was recently extended to include randomness and imperfect information. However, determining the extent to which this enhancement allows to describe truly arbitrary games was left as an open problem. We provide a positive answer to this question by relating the extended Game Description Language to the universal, mathematical concept of extensive-form games, proving that indeed just any such game can be described faithfully.

A contemporary grand AI challenge, General Game Playing is concerned with systems that can understand the rules of arbitrary games and learn to play these games well without human intervention. This principle has the great potential to bring to a new level artificially intelligent systems in other application areas as well. In this paper, our specific interest lies in General Trading Agents, which are able to understand the rules of unknown markets and then to actively participate in them without human intervention. To this end, we develop the existing Game Description Language (GDL) into a language that allows to formally describe arbitrary markets such that these specifications can be automatically processed by a computer. We present both syntax and a transition-based semantics for this Market Specification Language and illustrate its expressive power by presenting axiomatizations of several well-known auction types. 1

General Game Playing (GGP) research aims at building automated intelligent computer agents that accept declarative descriptions of arbitrary games at run time and are able to use such descriptions to play effectively without human intervention. This paper extends the Game Description Language (GDL) to more easily describe complicated games, including social and financial games with diverse goals. The GDL extension allows adding functions, defined in other programming languages such as Java, C++, Lisp, to the general game playing architecture. We used the three-player, simultaneous financial game Farmer, a standard GGP game, as an example to illustrate the need for and the benefits of the GDL extension. We also show the extension of the GDL is critical for adding coalition games to the general game playing area. 1