An architecture for decision-based processor

A main problem in building intelligent agents, as well as agent-based software, is testing since assuming a set of test cases is not possible. To solve this problem we introduce a modeling technique based on a formal representation that allows the verification of requirements. Based on this formal model we suggest a software processor that handles definition, storage, interpretation, and processing of decisions defined by an expert in the field and processed and used by an agent or a less experienced user. The processor has a built-in verifier that checks on completeness and validity of decisions. The architecture is based on the assumption that building an agent has to coincide with the structure of decision-making within the problem. The architecture adds a level of easy and dynamic to the agent environment.