| This thesis presents a tool, the Game Analyzer, which tests video games rapidly and provides useful feedback about the game's behavior to the game developer, who can then make adjustments to the game as desired. With the help of user-defined abstract states and actions, the Game Analyzer builds a state-transition model for a user-defined scenario within the game. This state-transition model defines the set of possible policies (gameplay strategies). The Game Analyzer then evaluates a large sample of the policies. These results can then be manipulated and visualized to help the game developer fine tune the scenarios played in the game and the game engine itself. This research was applied mainly to Role-Playing Games but could be applied to other domains. |
