Simulated competitions to aid tactical skill acquisition

The paper presents a framework to support human skill acquisition of game tactics (e.g., chess playing). We argue that cooperative work for defining heuristic constituents carried out by a learner should be alternated with simulated competitions to provide a formal, rating based feedback during the training phase. Firstly, our general definition of tactic concepts is bound to heuristic knowledge formalisations of two-player board games, including the notions of temporal, positional and material advantages. Secondly, our tactics definition language, aimed at the trainee, is described to cover a wide range of semantic features that can be applied in artificial games through a minimax search-based engine. The definition of heuristic parameters is based on variations of quantitative and qualitative production rules. The framework is instantiated by implemented software tools for the domain of chess. Finally, we draw conclusions about the suitability of the claims based on an empirical study.