Amirkabir University of Technology

Computer Engineering & IT Faculty

Thesis Title: Improvement of an Agent Oriented Methodology : MaSE

Student: Shiva Vafadar

Supervisor: Dr.Ahmad Abdollahzadeh

Abstract

This thesis defines a new methodology in agent oriented software engineering. Ex-MaSE (Extended Multiagent System Engineering) methodology is an extension of an existing methodology, MaSE. By using MaSE to implement a case study in electronic commerce, weaknesses of MaSE methodology in different phases of system life cycle have been reported. In order to bridge the gaps in original methodology, a new methodology named Ex-MaSE has been introduced which has “Environment Modeling”, “Knowledge Modeling” and “Agent-Object Modeling” steps in addition to original methodology steps.

“Environment Modeling” step has been added to the analysis phase and contains two steps “Capturing Environment Elements“and “Identifying Environment characteristics”. Purpose in this step is identifying environment entities of the system and properties of each role and entire of the multiagent system.

“Knowledge Model” has been proposed for modeling knowledge of each role in the analysis phase. The system analyst should specify required knowledge for each role to achieve its associated goals and to handle complexity caused by each environment type. “Fact Schema”, “Plan Schema” and “Rule Schema” are models of this step.

A role that has a deterministic response to messages and does not have the desired characteristics of an agent is essentially a simple object in an agent-based system. Because objects are simpler computational entities than software agents in such cases it is more efficient to use object abstraction. Therefore, “Agent-Object Model” has been introduced in the new methodology. In this model designer can map a role to either an agent or an object.

Using these improvements Ex-MaSE became more expressive, refinable, accessible and extendible than the MaSE methodology. Also reusability and maintainability of the developed systems using the Ex-MaSE methodology will improve.

Keywords: Agent oriented Software Engineering, Agent oriented methodology, MaSE