Modeling expert players´ behavior through data mining

Games are a field in which players are implicitly trained and compelled to solve hard problems optimally. By making an appropriate training experiment using a computer game, players can reach expert performance. A computer game tournament that contains Euclidian traveling salesman problem (ETSP) instances is developed and solutions made by expert players are analyzed with human-behavior hypotheses and data mining techniques. A model for the expert players´ behavior that combines apparently different fields is proposed and tested with TSPLIB problems.