摘要

In this contribution, I attempt to improve upon my existing computational model for recognizing beauty in mate-in-3 combinations in the game of international (or Western) chess. The intention is to obtain sonic insight into the way the existing model may be applicable outside the current scope, e.g., to single moves and endgame studies. The full article consists of two parts. The first part contains two phases of experimentation which compare combinations taken from the domain of compositions and from real games. In both phases I use a yardstick of human-player aesthetic ratings. In this part, we report three results. First, it was discovered that only having a high positive correlation with the human rating does not necessarily mean that (this variation of) the model is viable. Second, variations of the existing model in terms of the aesthetic features examined and the weights attributed to them are demonstrably either worse or, in the minority of cases examined, at best equivalent in performance to it. So, my original model may, at this moment, be adequate. Third, experimental results lead to questions on the effectiveness of using different weights (even those provided by domain experts) with respect to aesthetic features for the purpose of discriminating between them in terms of inherent 'importance'. In practice, any discriminating procedure was found to be unreliable and therefore it offered no improvement over the default intelligently designed feature evaluation functions that, in principle, do not value some features over others.

  • 出版日期2010-9