Action videogames are dynamic, complex environments, quite challenging to model. A research objective of HILab is the application of various modeling techniques to such environments with the goal to evaluate and compare their performance on the estimation of the similarity in the playing techniques of various players.

The dataset for modeling Frogger (http://www.frogger.net/), in order to classify players according to their playing technique (aggressive, defensive etc), is available for experimentation. Please cite HILab when using the data.

The datasets (the “holistic” and “grid” approach) for modeling the action game SpaceDebris are available for experimentation. The approach has been documented and published in the Workshop of Machine Learning and Data Mining in Games. Please cite the publication when using the data.

Costas Boletsis, Dimitra Chasanidou, Panagiotis Pandis, and Katia Lida Kermanidis. 2011. Semantic Representation of Action Games. Proceedings of the ECML/PKDD Workshop of Machine Learning and Data Mining in Games. Athens, Greece, September 9, 2011.

SpaceDebris has also been modeled by applying Latent Semantic Analysis (LSA). A zip file containing the LSA-reduced term-document matrices with the holistic, the grid, and the merged grid approach for three different dimensionality reduction ratios are available for experimentation. The approach has been documented and published in the European Conference on Artificial Intelligence. Please cite the publication when using the data. (The data used in the publication are only a subset of the data on this link.)

Katia Lida Kermanidis, Panagiotis Pandis, Costas Boletsis, Dimitra Chasanidou. 2012. LSA for Mining Hidden Information in Action Game Semantics. Proceedings of the 20th European Conference on Artificial Intelligence, 27–31 August 2012, Montpellier, France – Including Prestigious Applications of Artificial Intelligence (PAIS-2012) System Demonstrations Track