ePrints.FRI - University of Ljubljana, Faculty of Computer and Information Science

Computer analysis and comparison of chess players' game-playing styles

Urša Krevs (2015) Computer analysis and comparison of chess players' game-playing styles. MSc thesis.

Download (4Mb)


    Today's computer chess programs are very good at evaluating chess positions. Research has shown that we can rank chess players by the quality of their game play, using a computer chess program. In the master's thesis Computer analysis and comparison of chess players' game-playing styles, we focus on the content analysis of chess games using a computer chess program's evaluation and attributes we determined for each individual position. We defined meaningful attributes that can be used for computer analysis and are also comprehensible to a chess player. Using the attributes, we built profiles with which we defined chess players' styles. We evaluated the quality of the profiles by automatically identifying a chess player in a set of chess players. The profile of the chess player in the set and the one outside of it was built using different chess games. Using the result of the analysis we refined the profiles structure. In doing so we were aided by the information gain of each attribute. The most suitable profile was used for searching for world chess champions with the most similar style to a chosen chess player. We also sorted the world chess champions into groups according to their style of play. Because these players are well known, we compared our groups with the chess players' actual styles and determined how successful we were. By using the developed profiles, we can help partly automate and ease a chess player's analysis of chess games. We believe that the methods used in building the profiles and for the subsequent analysis could be applied to other domains.

    Item Type: Thesis (MSc thesis)
    Keywords: chess, computer analysis of chess games, playing style, group discovery
    Number of Pages: 87
    Language of Content: Slovenian
    Mentor / Comentors:
    Name and SurnameIDFunction
    doc. dr. Matej Guid937Mentor
    Link to COBISS: http://www.cobiss.si/scripts/cobiss?command=search&base=51012&select=(ID=1536276931)
    Institution: University of Ljubljana
    Department: Faculty of Computer and Information Science
    Item ID: 2973
    Date Deposited: 02 Apr 2015 17:33
    Last Modified: 16 Apr 2015 12:43
    URI: http://eprints.fri.uni-lj.si/id/eprint/2973

    Actions (login required)

    View Item