Search this site
Search WWW

This website is best viewed at 800x600 resolution.Internet Explorer 5 and Netscape6 are both recommended for better viewing result.

Copyright © 2001 University of Otago and its licencers. All rights reserved

About Software Agents
 

The notion of a software agent takes its inspiration from a human agent. Here is how AgentLink (p10 of reference below) defines a software agent:

Agents can be defined to be autonomous, problem-solving computational entities capable of effective operation in dynamic and open environments. Agents are often deployed in environments in which they interact, and maybe cooperate, with other agents (including both people and software) that have possibly conflicting aims. ... Agents can be distinguished from [software] objects ... in that they are autonomous entities capable of exercising choice over their actions and interactions.

Although agents are distinguished from objects, agent-based systems are commonly constructed using object-oriented technology. A key component to agent interaction is the use of knowledge-level communication technology that provides support for extended (over time) interactions among participating agents. This knowledge-level technology involves the use of formally defined ontologies and the symbolic modelling of possible interaction protocols between agents. In this context,

  • An ontology is an explicit representation of a problem domain that provides a high-level and standardised terminology for encoding exchanges of information and requests for services.
  • An interaction protocol is a description of an agreed protocol that a group of agents may use to exchange messages for a particular purpose, e.g. for initiating and responding to service requests, or to conduct more complicated interactions such as auctions.

Further development of agent interaction technology that we are pursuing involves

  • the use of agent mechanisms that refer to explicit representations of institutions (agent group interactions), agent commitments (in the context of an institutional setting), and policies (e.g. the establishment of agent security policies and software mechanisms for their enforcement)
  • the employment of narrative techniques for the guidance of long-term agent behaviour.

Some overviews of basic agent technology available on the web are

Articles about agent communication languages and ontologies: