Combining Answer Set Programs for Adaptive and Reactive Reasoning

In this paper we present a method to improve efficiency of ASP based system in dynamic context. Our proposal is a reasoning framework based on combination of ASP modules. This framework can be seen as a tool to assist the development of ASP based systems. Our final goal is to provide a general design pattern to help construction of efficient ASP based reasoning systems. In this paper, we first define a method to divide the wide program into different kinds of modules. Then, we show how meta-knowledge on these decomposition can be used for meta-reasoning. We discuss the benefit of this modular reasoning method through an example of multi-agent systems in a dynamic world. Using this example, we show that this method can elegantly allow agents to realise dynamic behaviour via introspective meta-reasoning. Finally, through experimental results we show that the framework can be used to improve efficiency of ASP based system in terms of reactivity and adaptability.