« Lord Of The Strings | Main | Formel för klackhöjd »
mars 19, 2004
Multi-agent technology samt lite om ontologier och annat
Artikeln Multi-agent technology: removing the 'artificial' from AI beskriver en av de grenar som bygger system kring agenter (se nedan för en begreppslig not).
AI technologies have been around for years, but they don't always work as well as we'd like. Anyone who has used speech-recognition technologies in a business context will know that the result can be, at best, laughable. And, should your voice be changed by something like a cold, the whole thing falls apart.
But technology has advanced since AI was first developed. Magenta Technology is one company that is developing a new type of technology that uses multi-agent systems to make informed decisions based on the resources that are available. Agents are small software programs that communicate with each other, acting behaviorally to interact and respond, matching available resources to demand.
Det förklaras kort även om relationen mellan dessa Multi-Agent System och ontologier.
These agents are able to perform in this way owing to their ability to communicate with a database based on ontology. Based on principles first described by Aristotle, ontology is that part of metaphysics that deals with the nature and essence of being or existence. In the context of multi-agent systems, ontology is a computer-readable description of knowledge about the resources in an enterprise's network. It relates attributes to different classes of objects, such as available business resources, projects, customer orders or plans, to enable knowledge of available resources to be built up in ontology. The software agents become intelligent because they can make use of the knowledge contained in ontology to use in the process of negotiation and decision-making.
...
At present, most work being done in the world of multi-agent technology is focused on logistics applications, but it also holds promise for the development of predictive systems in the financial services market as well as in the development of next-generation intelligent computer games. In an era where all companies are looking to increase their visibility into the workings of their supply and value chains, multi-agent technology holds enormous promise for making intelligent business decisions in real time in complex environments.
Via Agents Portal.
Notera att det finns en viss koppling mellan sådana "agenter" man talar om i denna Multi Agent Technology och agenter t.ex. i agentbaserad modellering som tidigare skrivits om här, men de bör inte förväxlas. En skillnad är att man i de senare inte inte talar så mycket om ontologier. Begreppet "agent" används för att beteckna många olika med snarlika saker (som nog alla har en familjelikhet för att prata Wittgensteinska).
Se även
Magenta Technology (uppdaterad länk)
UMBC AgentWeb
Ontology Development 101: A Guide to Creating Your First Ontology
John F Sowa: Ontology och Agent.
Recension: Michael Crichton 'Prey'.
En liten sidestep
Jag hittade även en sajt om och med Bishop John Wilkins An Essay Toward a Real Character and a Philosophical Language som bl. a. Umberto Eco kommenterat (se sidan för länkar). Jorge Luis Borges kritiserade boken i sin The Analytical Language of John Wilkins, och som det står på sajten: apparently done without having read it.
Se även Kategorierna och Borges.
Posted by hakank at mars 19, 2004 10:02 FM Posted to Diverse