主催: The Japanese Society for Artificial Intelligence
会議名: 2013年度人工知能学会全国大会(第27回)
回次: 27
開催地: 富山県富山市 富山国際会議場
開催日: 2013/06/04 - 2013/06/07
When designing and evaluating the performances of a system in terms of resilience, it is crucial to consider not only its global configuration, but also its dynamics with regard to the environment. Indeed, our systems are subject to uncontrollable events and their analysis requires to capture these events as well as the controlled events (i.e., the actions) in their global succession relationships. In this paper, we define a language that is expressive enough to represent any narrative scheme, that is, any set of total preorders over any finite set of events. We then formalize the problem of existence of an adequate ``strategy'' for a given narrative scheme, that consists in adding some actions between the uncontrollable events (leading to a specific scenario) in a way that the system satisfies some expected property. Then, each scenario is interpreted as a sequence of propositional formulae that are updated each time an uncontrollable event or an action occurs. Lastly, we propose some properties in order to characterize the notion resilience for such dynamic systems.