Meta-interpretation of the t-Sapi Language

Recently an effort in design and realization of artificial agents and multi-agent systems reached significant interest within information technology research community. One of topics that also support this progressive area is the modelling and simulation of the mult-agent systems. We are developing a tool for modelling and prototyping of such systems where the intelligent agents are driven by an original control language called t-Sapi. The language was proposed on a low level of abstraction to allow partly an easy realization of the agents at different platforms and partly building agents with based on different principles. But some complex agent architectures, as are the BDI-based ones, needs to be able make meta-reasoning (so there must be a multi-interpretation abilities) to handle failed plans, to build and select new intentions, or to make decisions about plan reconsideration. This was one of the reasons to think about of a t-Sapi meta-interpretation. Following article outlines its basic principles and shows how it could be used for creation of agents that uses meta-reasoning.