Planning an argument based on stochastic knowledge of the opponent. Given an initial optimal strategy and the opponent model(s) against which this strategy is guaranteed to succeed, the task is to undergo a replanning every time the computer runs out of strategies to assert in the dialogue game. If the replanning is successful, it generates a modified/extended strategy guaranteed to succeed against a larger subset of the opponent models as compared to the initial optimal strategy.

The files 'argstrat.py' and 'argstrat6.py' have been specially designed for cycle4.txt and cycle6.txt in the examples folder. 'argstrat.py' corresponds to 'cycle4.txt' and 'argstrat6.py' corresponds to 'cycle6.txt' In order to run the two python scripts, one needs an automated planner. Please contact Dr. Amanda Coles for the planner. (amander dot coles at kcl.ac.uk). For cycle 4, run

$ python argstrat.py ./planner ./examples/cycle4.txt

Follow the instructions as they appear in the terminal. If the replanning is successful, the terminal screen prompts the user to enter new model identity. Enter the numeric value of all the models against which the strategy generated by the planner is successful. For example, if the terminal shows 'a10', enter 10. Type 100 to quit. Next, the terminal screen prompts the user to enter an argument. Enter the argument which doesnot appear in the planner output. Press any numeric value if such an argument does not exist. Follow the same instructions for cycle6.

$ python argstrat6.py ./planner ./examples/cycle6.txt

P.S. Some files are borrowed from https://github.com/christopher-hampson/argstrat which solves a problem similiar in nature to the problem at hand.