FIPA Agent Communication Technologies and Services
While several groups have identified the need for agent standards, their production is currently a research activity. Although research groups in various companies have produced working multi-agent systems, many issues are at present too poorly understood for reliable, well-specified standards to be generated. The goal of this project is therefore to generate this understanding such that the project can act as a driver of relevant standardisation effort.
The principal focus of the project is the standards under development by FIPA - the Foundation for Intelligent Physical Agents. Additionally, the project will seek to influence agent standardisation work being performed by the Object Management Group (OMG).
The project will be structured around two development cycles - phase 1 and phase 2 - which will mirror the timetable for the production of agent standards in FIPA. Phase 1 will focus on the FIPA97 normative specifications, published in October 1997, with the aim of driving the development of FIPA98. Similarly, Phase 2 will validate the FIPA98 normative specifications due the following year, aiming to drive ongoing standardisation activity.
In each phase, demonstrators will be produced in three application areas - audio-visual broadcasting, service reservation, and electronic commerce. These application areas have been chosen to provide good coverage of the functionality of the FIPA and OMG standards.
Summary of Trial
The trials performed by the FACTS project will be designed to stress the full scope of standards under development by FIPA. At this stage, is has not been decided whether the three software interaction trials will each be performed on a single site or by linking remote sites.
The principal challenge for the project is to show that agent-based systems produced independently by different partners can achieve interoperability by conforming to emerging standards. It is fully expected that weaknesses will be found in current standards that prevent achievement of this goal, which will require the project to evolve 'patches' and to lobby the relevant standards activity to implement these in upcoming standards revisions.
The project will demonstrate the feasibility of producing complex agent systems using components developed by different companies and will therefore prove the feasibility of open agent-based trading environments.
It is also expected that the project will prove a major stimulant to the development of agent standards and that it will establish strong skills and competencies in agent technology in the partner companies.
We expect that agent technologies will be key to many future electronic commerce applications - performing searching, filtering, and negotiation tasks on the user's behalf.
Elsewhere, agents will act as 'personal assistants' to the user, hiding the underlying complexity of future systems, helping to make advanced services more usable.
However, such future applications will demand the existence of open agent environments - 'marketplaces' and working environments where agents can interact with one another.
The FACTS project will be instrumental in the enabling such environments.