Story-based planning in theme park

This study develops a planning system to make a tour plan both in a theme park and a town with multi-agent architecture. The main part of the system comprises two kinds of agents: a Story Writer agent and a Story Miner agent. A Story Writer agent makes a schedule for a user based on the user's preferences, goals, and interests with a consistent plot such as ride all the coasters in the park or go to see a trendy movie. It contributes to a user's satisfaction by generating an interpretable plan that matches the plot. On the other hand, the Story Miner agent discovers some characteristic from the generated plan and produces an explanation for why the plan is good. It contributes to user satisfaction by explicitly providing the reason for the plan being good. We describe two scenarios: a theme park scenario and a date support scenario. Especially, we detail algorithms for both types of agents for theme park scenario.