Make your own free website on Tripod.com







[an error occurred while processing this directive]



Introduction

Understanding natural language, that we could call "speech language" too, is actually one of the arguments of greater interest in the artificial intelligence field. In order to analyze the problem, suppose we are trying to implement an intelligent agent in a position to use and understand natural language. The first question is: "what is an intelligent agent?". An intelligent agent, as general definition, is a something that interacts with the environment that encircles it. Perhaps what I said is a little too generic and it doesn't help us to clear ideas, however adding particulars to the last definition could carry us to make mistakes. Let's try to add something to our definition. Usually an intelligent agent is made up by an architecture (physical) through which the agent is in a position to interacting with the world (to complete actions and receives perceptions), and from a program that analyzes the perceptions received from the external world and indicates the actions to complete. But now we could ask "Why the agent must complete an action?". Well, an agent does not exist for case, if someone has "lost his time " to create it there must be a reason. I mean, an agent has always a objective to reach (which is generally called "goal") and it executes actions in order to catch up its scope. Let's have an example. An agent created to playing poker would have to be endowed of a television camera through which he can receive visual perceptions (as an example the cards played by the adversaries), a pair of mechanical arms in order to move cards from and to the table and so on. Well, this is the agent's architecture, but it is not enough a television camera and a pair of arms to become a good poker player! Our agent needs a program which suggest him the game strategy: which cards to change, when to raise, when to leave and so on (...I do not know much more about poker strategy because I've never won one single game). We could continue to speak about our poker player, or however about intelligent agents, to distinguish them in classes according to their way of learning, planning the actions, representing the world that encircles them... but it is not about this that we want to discuss in this moment.




Italia HyperBanner Network