From AOSE methodology to agent implementation

  • Authors:

  • Philippe Massonet;Yves Deville;Cédric Nève

  • Affiliations:

  • CEDITI, GOSSELIES, Belgium;Université catholique de Louvain, LOUVAIN-LA-NEUVE, Belgium;CEDITI, GOSSELIES, Belgium

  • Venue:
  • Proceedings of the first international joint conference on Autonomous agents and multiagent systems: part 1
  • Year:
  • 2002

Quantified Score

Hi-index 0.00

Visualization

Abstract

This article investigates the transition between semi-formal AOSE methodologies and FIPA compliant agent implementation tools. It focuses on the translation between an implementation independent design and an implementation dependent detailed design/ implementation. It is illustrated on a case study using a subset of the MESSAGE methodology whose modelling language extends the UML, the JADE development framework and the Jess rule based system for reasoning. It shows how the implementation is partially structured based on the design description where high-level design decisions have been made.The case study suggests that agreement on a minimal set of standard high-level agent concepts for AOSE modelling is needed. This would enable development tools to support these standard agent concepts. The current situation where development tools have their own set of basic agent concepts implies that the translation process is unique for every combination of methodology and development tool. This hinders the ability to define a complete development process from analysis to design and implementation and is an obstacle to the adoption of the agent approach by mainstream software engineering departments.