Um modelo genérico de agente racional baseado em raciocínio assíncrono interativo.

Abstract

The ϕ-calculus theoretical tool tries to manage asynchronous interactive reasoning in the empirical sciences context. An instructor (a human specialist agent of the domain) with knowledge about the subject and interest in reviewing and evolving this knowledge collaborates with artificial agent which learns (and reasons about) the knowledge that the instructor withholds. It is important to verify how such knowledge evolved, so after the learning phase, two other phases initiate: the knowledge evaluation (by the domain specialist) and the correction (which happens after the refining of the learnt knowledge). This collaboration/learning process between the domain specialist and the artificial agent is interactive and asynchronous. An architecture, inspired in ϕ-calculus and based on interactive asynchronous reasoning, is proposed.