Revisions of orders in dynamic systems

Abstract

Current reasoning systems attempt to model an agent's knowledge and interaction with its environment in a symbolic manner. This environment, its world is generally dynamic and changing due to natural evolution or the actions of other agents that are a part of it. In consequence, an agent that is a part of a reasoning system must have the following components: a knowledge base where its knowledge of the world is stored, a communication mechanism with the environment and other agents in it, and a means of modifying its knowledge of the environment. Knowledge may be represented by a logic language which is propositional, first order, modal or extentions of these. Each one of these alternatives has advantages as well as disadvantages. The higher the expressive power of a given language, the more computational problems there are regarding complexity and decidability. Communication mechanisms can be varied, depending on the environment being modeled. They can be multimedia mechanisms such as microphones, speakers, video cameras, infrared sensors, motion detectors and even wired or wireless systems where information is transmitted without any kind of preprocessing. They are irrelevant, however, for the purpose of our research because we are focused in the development of the knowledge system. Mechanisms for modifying knowledge may be modeled by what is known as Belief Change Theory. Belief Change Theory assumes that the underlying language is at least propositional. An agent's knowledge is represented as a set of sentences and new information as a single sentence. In turn, every change operator takes a set of sentences and a single sentence and produces a new set of sentences as a result.