The goal of this group is to develop a framework for logging information from games in a common format that captures common-currency metrics like win/loss, score as a function of time, entropy measures on games state, and others listed subsequently. The framework provides an implementation of a number of general measures that have previously been used to describe gameplay in an abstract form. The relevant information from the game has to be extracted with game-specific code implemented by the user and can then be processed by our framework. The framework is thus capable of logging full game states at each tick of the game, but also allows users to analyse specific characteristics of gameplay. The framework will be made available at [https://github.com/GAIGResearch/GameEvaluationMetricsAtDagstuhl] and is intended to be extensible to include more measures. During the seminar, we have implemented a number of features within the framework and applied it to the GVGAI software, which already resulted in interesting observations.
Github: https://github.com/GAIGResearch/GameEvaluationMetricsAtDagstuhl
Cite this work
@inproceedings{volz17471, author= {Vanessa Volz and Dan Ashlock and Simon Colton and Steve Dahlskog and Jialin Liu and Simon M. Lucas and Diego Perez Liebana and Tommy Thompson}, title= {{Gameplay Evaluation Measures}}, year= {2018}, journal= {{Artificial and Computational Intelligence in Games: AI-Driven Game Design (Dagstuhl Seminar 17471)}}, volume= {7}, number= {11}, pages= {86--129}, editor= {Pieter Spronck and Elisabeth Andre and Michael Cook and Mike Preuss}, abstract= {The goal of this group is to develop a framework for logging information from games in a common format that captures common-currency metrics like win/loss, score as a function of time, entropy measures on games state, and others listed subsequently. The framework provides an implementation of a number of general measures that have previously been used to describe gameplay in an abstract form. The relevant information from the game has to be extracted with game-specific code implemented by the user and can then be processed by our framework. The framework is thus capable of logging full game states at each tick of the game, but also allows users to analyse specific characteristics of gameplay. The framework will be made available at [https://github.com/GAIGResearch/GameEvaluationMetricsAtDagstuhl] and is intended to be extensible to include more measures. During the seminar, we have implemented a number of features within the framework and applied it to the GVGAI software, which already resulted in interesting observations.},
}