i*


an agent-oriented modelling framework   

 
 
The i* framework proposes an agent-oriented approach to requirements engineering centering on the intentional characteristics of the agent.  Agents attribute intentional properties (such as goals, beliefs, abilities, commitments) to each other and reason about strategic relationships.  Dependencies between agents give rise to opportunities as well as vulnerabilities.  Networks of dependencies are analyzed using a qualitative reasoning approach.  Agents consider alternative configurations of dependencies to assess their strategic positioning in a social context.

The framework is used in contexts in which there are multiple parties (or autonomous units) with strategic interests which may be reinforcing or conflicting in relation to each other.  Examples of such contexts include: business process redesign, business redesign, information systems requirements engineering, analyzing the social embedding of information technology, and the design of agent-based software systems.

The name i* (pronounced eye-star) refers to the concept of distributed intentionality. 


Keywords: strategic relationships, intentional dependencies, multi-agent systems, autonomous agents, social networks, requirements engineering

                                                                                                                                                                                        
i* 2005
A One-Day Symposium and Two-Day Research Workshop

Road map for readings on i*
For a quick overview, you may start with RE97 which presents i* concepts in a Requirements Engineering context.

A vision for strategic actors modelling in the context of Agent-Oriented Software Engineering is outlined in AOSE01. A similar overview for the Information Systems context is presented in WIj01.    

For a simple presentation of the framework in a Business Process Redesign context, you may start with either WITS94 or IEEE Expert 96.  Both provide an overview of the framework, with the latter including an example of ConGolog process modelling.  The COOCS93 paper focuses on the Strategic Dependency model (called the Actor Dependency model at the time). These papers should be enough to get you started in doing i* modelling.

CACM99 provides an overview of goal-oriented analysis, which is an important part of i*.  An earlier illustration of i* reasoning appeared in HICSS94 (This is now part of  the Strategic Rationale model.)

The ICSE94 paper is an application of the i* framework to Software Processes and Organizations. It has examples of the distinction among Agents, Roles, and Positions, and a discussion of enforcement, assurance, and insurance.  These are further illustrated in ICEIMT97.

IFAC99 offers examples of Business Modelling, with an example about IKEA.  These are further illustrated in ER01.  The latter also shows how i* modelling can be used to help reason about Patents in a strategic business context, with examples from e-commerce (online travel booking). The mechanisms for modelling patent protection is similar to those for modelling Trust and Security in Trust00.

The WIj01 paper provides a more detailed explanation of the premises behind the i* ontology.  The main arguments are summarized in the AOIS01 invited presentation.

COOCS95 compares several different approaches to modelling organizational work, including IBIS, KAOS, and Action Workflow (speech acts).
IWSSD98 and CoopIS95 provide examples of how i* may be used in conjunction with other modelling or specification languages in the context of software development.

The ER94 paper contains some partial schemas and axioms for the models.  More details are available in the PhD thesis [Yu95].
 


A set of Powerpoint slides assembles presentations slides from various papers and talks into a tutorial format.


Tools

OME3
OME2


See also GRL - an elaborated version of i* , which is part of a proposal for a User Requirements Notation.


The i* framework is the basis and starting point for the Tropos project towards methodologies for agent-based systems.  Several research groups are involved in this work.    
Tropos at the University of Toronto, Canada
Tropos at the University of Trento, Italy
Tropos at the Technical University of Aachen (RWTH), Germany
Tropos at the Federal University of Pernambuco  (UPFE), Brazil

New central site :  www.troposproject.org

Some applications, extensions, or adaptations of i* :    (See also E. Yu Publications)
in selected areas:

Requirements Engineering

Process Analysis and Design, Reengineering Evaluation, Verification and Validation  (See also the Tropos project.) Agent-Oriented Systems Development  (See also the Tropos project.)


Trust in multi-agent systems

Security requirements engineering
Software engineering processes and organizations Data Management Processes Knowledge management
Systems and Organizational Architecture
Business modelling Intellectual Property Management
     

Goal-Oriented Requirements Engineering

The  i*  agent-oriented approach builds on and extends Goal-Oriented Requirements Engineering (GORE)  [REFSQ98].  In particular the softgoal concept in i* draws on techniques for dealing with non-functional requirements in software engineering, as presented in
Some other related GORE frameworks include:

For discussions of goal-oriented requirements engineering, see  [Kavlaki02] [Rolland03] [vanLamsweerde00


Requirements Engineering

For an overview of Requirements Engineering, see:

Further Readings
Some recent papers which mention or discuss i*:     Some recent papers concerning or mentioning the NFR framework:

A number of Ph.D. dissertations have made use of i* modelling:



back to  University of Toronto :: KM lab  ::  Eric Yu  
To see who has visited this page, click here
Nedstat Basic - Free web site statistics
This page last modified on: December 29, 2004
Previous update: October 13, 2003