Next: References Up: Revisiting pjawns: The comprehension, Previous: Abstract Contents
The AI will develop the current AI already present in one of the models. The intention is to develop the AI without referencing the model which already has a strong AI component. The plan is to develop, a system which scores the possible moves and then takes the highest scoring move. I think flexibility in the system would be modifiers in the scoring. For instance a more defensive player would favour score defensive moves higher than offensive moves, and an offensive player would do the opposite. Also the base scores could be changed, which are used by the neutral player. The AI will also utilise a look-ahead strategy and the depth of the search to be used can be modified. In a way this can be seen as a difficulty setting maybe from easy to normal. It is important to remember that the aim is not to create an unbeatable AI player but to have an experience of extending an existing EM model. The target-standard of the AI is an AI capable of beating the model with AI already implemented.
The interface component will allow the two specified models to play against one another. This will work by adopting the GUI of one of the models (model A) and have model B interact with model A as a human player and this will work the other way too. While there will be no alterations to any GUIs, information will be passed to the user through the terminal window. The interface will disable all board interactions that the user can have with the board, so that there are never two (potentially conflicting) inputs. However all other functionalities will remain, for instance to start a new game and enabling/disabling the AI of model A.