Agent Oriented Software Engineering Methodologies - CiteSeerX

Agent Oriented Software Engineering Methodologies Andrea Omicini Alma Mater Studiorum – Universit` a di Bologna [email protected]

Panel Agent Oriented Methodologies and Programming Languages Towards Practical Systems AAMAS 2009, Budapest, Hungary 13 May 2009

Andrea Omicini (UNIBO)

AOSE Methodologies

Panel @ AAMAS 2009

1 / 18

Outline

1

Why We Do Not Need To Worry About AOSE

2

Why We DO Need To Worry About AOSE

3

What We Should Think / Do About AOSE

4

Bibliography


AOSE Methodologies

Panel @ AAMAS 2009

2 / 18

Outline

1


2


3


4

Bibliography


AOSE Methodologies

Panel @ AAMAS 2009

3 / 18

It Works MAS represent indeed a novel general-purpose paradigm for software development Agent-based computing promotes designing and developing applications in terms of I I I

autonomous software entities situated in an environment that can flexibly achieve their goals by interacting with one another in terms of high-level protocols and languages

These features are well suited to tackle the complexity of developing software in modern scenarios [Zambonelli and Omicini, 2004] I

I

I

the autonomy of application components reflects the intrinsically decentralised nature of modern distributed systems, and can be considered as the natural extension to the notions of modularity and encapsulation for systems that are owned by different stakeholders the flexible way in which agents operate and interact (both with each other and with the environment) is suited to the dynamic and unpredictable scenarios where software is expected to operate the concept of agency provides for a unified view of artificial intelligence (AI) results and achievements, by making agents and MAS act as sound and manageable repositories of intelligent behaviours


AOSE Methodologies

Panel @ AAMAS 2009

4 / 18

It is the Only Thing with Some Chances to Work

So-called “traditional” approaches to SE have already shown their limits I

I

I

“Failure” is still a typical result of many large-scale software projects in the real world Costs are clearly too high: Small-medium enterprises cannot really afford neither in-house nor external software development for their specific, pressing needs In practice, time is always a problem in complex application scenarios

Aspect-oriented programming / methodologies? I I

quite successful not enough: there is not enough “aspectdness” to really face the complexity of pervasive systems, for instance


AOSE Methodologies

Panel @ AAMAS 2009

5 / 18

The Issue of Meta-Model The fact is, that only AOSE methodologies (and more generally, AO techniques) have the richness to face complexity at the right level of expressiveness I I I I I

Agent / Society / Environment Agents & Artifacts Intelligent agent architectures Organisational / Normative views on MAS ...

Many meta-models, still AOSE makes a uniform coherent view over complex artificial systems available I I I I I I I I

complex distributed systems intelligent systems service-oriented systems robotics systems pervasive systems adaptive & self-* systems socio-technical systems ...

Objects? Aspects? Services? . . . ??? I

Please, be serious


AOSE Methodologies

Panel @ AAMAS 2009

6 / 18

The Issue of Process

AOSE methodologies are the only viable approach to the engineering of socio-technical systems I

where the distinction between human and artificial actors first blurs then fades away

AOSE researchers are running along the path of meta-modelling to build a coherent and effective model of SE processes I

new “Design Process Documentation and Fragmentation” IEEE FIPA Working Group coming next

The SE Process is a Socio-technical Systems I

the AOSE paradigm is the only one capable of directly expressing both the system and the process in the same uniform way


AOSE Methodologies

Panel @ AAMAS 2009

7 / 18

The State of the Art Research on AOSE is alive [Henderson-Sellers and Giorgini, 2005, Bergenti et al., 2004] I I

I I I

I

AOSE methodologies still under development, toward mature state developing lines of research on meta-modelling, fragmentation & method composition, on effective AO tools & frameworks, on the bottom-up approaches emerging from leading AO technologies. . . the IJAOSE journal seems to have finally achieved a good standard some funded projects on pure AOSE themes – e.g., “MEnSA” in Italy a lot of cooperation among the AOSE researchers, in Europe in particular, but not limited to AOSE TFG’s last meeting was in Bath, UK, last December, even without funding [Bernon et al., 2005]

Teaching of AOSE is gaining momentum [Luck et al., 2004] I

A number of MAS courses spend some time on AOSE methodologies, at least in European universities


AOSE Methodologies

Panel @ AAMAS 2009

8 / 18

Outline

1


2


3


4

Bibliography


AOSE Methodologies

Panel @ AAMAS 2009

9 / 18

Issues & Problems AOSE methodologies are no longer perceveid as a cooooool topic by many researchers → not enough research on new AOSE issues Tools are not perceived by many researchers as essential elements of their research → not enough work on effective tools Too few projects on the specific issues → not enough money invested Slow penetration in the industry → not enough practice & experience for AOSE methodologies to mature Teaching methodologies is boresome → not enough teaching for AOSE methodologies to spread fast


AOSE Methodologies

Panel @ AAMAS 2009

10 / 18

Outline

1


2


3


4

Bibliography


AOSE Methodologies

Panel @ AAMAS 2009

11 / 18

What Should We Think about AOSE? Think AOSE methodologies are boring but cool I

I

they are the only hope for the definition of a solid engineering process for artificial systems of tomorrow just think of artificial systems instead of simple software ones

AOSE methodologies should aim at artificial systems instead of mere software ones I

they could work as the basis for a general discipline of system engineering

AO technologies & AOSE frameworks should be aimed at making the costs of paradigm shift as low as possible I

they should promote coherent and well-founded blending of OO technologies and models in the MAS conceptual context


AOSE Methodologies

Panel @ AAMAS 2009

12 / 18

What Should We Do about AOSE?

Do Using AOSE methodologies in projects and complex application problems Bridging between “traditional” and AO software engineering techniques and notations Driving the development of AOSE tools within projects and advanced applications Teaching AOSE methodologies systematically in SE courses, linked to non-trivial student projects


AOSE Methodologies

Panel @ AAMAS 2009

13 / 18

Conclusion

So What? Keep on researching in AOSE – so much still to do & achieve Do not fret about AOSE – so much time still ahead of us Look back carefully – most of what we learned about SE is likely to be of some use in AOSE Look forward with an open mind – most of the systems to be engineered in the near future have no answer from traditional SE theory and practice, new ideas required But mostly, take care of AO technologies, which will be the key to the widespread practice for AO-whatever


AOSE Methodologies

Panel @ AAMAS 2009

14 / 18

Outline

1


2


3


4

Bibliography


AOSE Methodologies

Panel @ AAMAS 2009

15 / 18

Bibliography I Bergenti, F., Gleizes, M.-P., and Zambonelli, F., editors (2004). Methodologies and Software Engineering for Agent Systems: The Agent-Oriented Software Engineering Handbook. Kluwer Academic Publishers. Bernon, C., Cossentino, M., and Pav´ on, J. (2005). An overview of current trends in european AOSE research. In [Omicini et al., 2005], pages 379–390. Henderson-Sellers, B. and Giorgini, P., editors (2005). Agent Oriented Methodologies. Idea Group Publishing, Hershey, PA, USA. Luck, M., Ashri, R., and d’Inverno, M. (2004). Agent-Based Software Development (Agent-Oriented Systems). Artech House Publishers.


AOSE Methodologies

Panel @ AAMAS 2009

16 / 18

Bibliography II

Omicini, A., Petta, P., and Gams, M. (2005). Special issue: Hot topics in european agent research I. Informatica, 29(4):377–443. Zambonelli, F. and Omicini, A. (2004). Challenges and research directions in agent-oriented software engineering. Autonomous Agents and Multi-Agent Systems, 9(3):253–283. Special Issue: Challenges for Agent-Based Computing.


AOSE Methodologies

Panel @ AAMAS 2009

17 / 18

Agent Oriented Software Engineering Methodologies Andrea Omicini Alma Mater Studiorum – Universit` a di Bologna [email protected]

Panel Agent Oriented Methodologies and Programming Languages Towards Practical Systems AAMAS 2009, Budapest, Hungary 13 May 2009


AOSE Methodologies

Panel @ AAMAS 2009

18 / 18