Human-like Heuristics in Scrabble

  • Authors:

  • Alejandro Gonzalez Romero;René Alquezar;Arturo Ramirez;Francisco Gonzalez Acuòa;Ian Garcia Olmedo

  • Affiliations:

  • Dep. de Llenguatges i Sistemes Informàtics, Universitat Politècnica de Catalunya;Dep. de Llenguatges i Sistemes Informàtics, Universitat Politècnica de Catalunya;Centra de Investigación en Matemáticas, A.C. (CIMAT), Guanajuato --México;Centra de Investigación en Matemáticas, A.C. (CIMAT), Guanajuato --México and Institute de Matemáticas, Universidad National Autónoma de México;-

  • Venue:
  • Proceedings of the 2009 conference on Artificial Intelligence Research and Development: Proceedings of the 12th International Conference of the Catalan Association for Artificial Intelligence
  • Year:
  • 2009

  • World-championship-caliber Scrabble

    Artificial Intelligence - Chips challenging champions: games, computers and Artificial Intelligence

  • Building Policies for Scrabble

    Proceedings of the 2008 conference on Artificial Intelligence Research and Development: Proceedings of the 11th International Conference of the Catalan Association for Artificial Intelligence

  • Opponent modeling in scrabble

    IJCAI'07 Proceedings of the 20th international joint conference on Artifical intelligence

Quantified Score

Hi-index 0.00

Visualization

Abstract

The game of Scrabble has been tackled by computers mainly using simulation [1--4]. This technique has already given very positive results. Nevertheless it fails to resemble human approach of thinking when playing Scrabble. A more human-like approach would be desirable both to test human strategies vs. simulation-based computer strategies and to improve and check the soundness of human strategies. This type of approach is being developed with good results [5], yet there is still much room for improvements in this direction. The current approach uses only one heuristic function for the whole game. Playing several games using this heuristic has suggested that other heuristics for the late phases of the game would be convenient. In this paper we recall a second heuristic from [6] and propose a third heuristic for the late phases of the game (Pre-endgame and endgame). We show examples of positions in which these heuristics apply and finally we test their performances using a pool of 1000 random games.