Building an Agent Methodology from Fragments: the MEnSA experience - - PowerPoint PPT Presentation

Building an Agent Methodology from Fragments: the MEnSA experience

Mariachiara Puviani & Massimo Cossentino Giacomo Cabri & Ambra Molesini

Dipartimento di Ingegneria dell’Informazione, Universit` a di Modena e Reggio Emilia, ICAR, Consiglio Nazionale delle Ricerche – Palermo, Alma Mater Studiorum—Universit` a di Bologna

AOMIP@SAC 2010, Sierre, Switzerland, 25th March 2010

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 1 / 21

1

The MEnSA Process Requirements

2

The New MEnSA Process

3

Results Assessment

4

Conclusions and Future Works

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 2 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . ◮ . . . but to reuse “fragments” of existing methodologies (PASSI, Tropos,

Gaia and SODA) by composing them through the Situational Method Engineering approach

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . ◮ . . . but to reuse “fragments” of existing methodologies (PASSI, Tropos,

Gaia and SODA) by composing them through the Situational Method Engineering approach

In order to do this we

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . ◮ . . . but to reuse “fragments” of existing methodologies (PASSI, Tropos,

Gaia and SODA) by composing them through the Situational Method Engineering approach

In order to do this we

◮ define the methodology’s requirements Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . ◮ . . . but to reuse “fragments” of existing methodologies (PASSI, Tropos,

Gaia and SODA) by composing them through the Situational Method Engineering approach

In order to do this we

◮ define the methodology’s requirements ◮ select the more suitable fragments Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

The objective of this paper

Our work is aimed at building a new methodology in order to fill the existing gap between agent-oriented methodologies and MAS infrastructures This was the objective of the MEnSA (Methodologies for the Engineering of complex software Systems: Agent-based approach) project, where we studied how to build a new methodology that takes into consideration the infrastructures’ features

◮ MEnSA aim was not to create a new-brand methodology. . . ◮ . . . but to reuse “fragments” of existing methodologies (PASSI, Tropos,

Gaia and SODA) by composing them through the Situational Method Engineering approach

In order to do this we

◮ define the methodology’s requirements ◮ select the more suitable fragments ◮ assemble fragments for creating a new methodology Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 3 / 21

Situational method engineering

Each methodology can be decomposed into reusable method fragments A designer can re-use and re-assemble fragments in order to create a new methodology [Cossentino et al., 2007] First step: extraction and storing of method fragments in the method base Second step: selection of the suitable fragments from the method base Third step: fragments assembly

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 4 / 21

Outline

1

The MEnSA Process Requirements

2

The New MEnSA Process

3

Results Assessment

4

Conclusions and Future Works

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 5 / 21

Initial requirements

1 To fill the gap between design and implementation: ◮ a support for traceability ◮ the abstractions adopted in the design phase should be ⋆ powerful enough for properly design the system ⋆ “near” to the abstractions supported by MAS infrastructures Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 6 / 21

Initial requirements

1 To fill the gap between design and implementation: ◮ a support for traceability ◮ the abstractions adopted in the design phase should be ⋆ powerful enough for properly design the system ⋆ “near” to the abstractions supported by MAS infrastructures 2 To adopt a complete requirements analysis phase Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 6 / 21

Initial requirements

1 To fill the gap between design and implementation: ◮ a support for traceability ◮ the abstractions adopted in the design phase should be ⋆ powerful enough for properly design the system ⋆ “near” to the abstractions supported by MAS infrastructures 2 To adopt a complete requirements analysis phase 3 To adopt proper levels of abstraction in order to deal with complex

problems

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 6 / 21

Initial requirements

1 To fill the gap between design and implementation: ◮ a support for traceability ◮ the abstractions adopted in the design phase should be ⋆ powerful enough for properly design the system ⋆ “near” to the abstractions supported by MAS infrastructures 2 To adopt a complete requirements analysis phase 3 To adopt proper levels of abstraction in order to deal with complex

problems

4 To enable an easy transition towards the new methodology to

designers fluent with one or more of the “source” methodologies

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 6 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge 10 FIPA-compliance is advisable at least at the communication level Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge 10 FIPA-compliance is advisable at least at the communication level 11 Gaia’s organisational rules: interesting approach for modelling some

social aspects

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge 10 FIPA-compliance is advisable at least at the communication level 11 Gaia’s organisational rules: interesting approach for modelling some

social aspects

12 Modelling the environment is important and could be done by

adopting abstractions from SODA

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge 10 FIPA-compliance is advisable at least at the communication level 11 Gaia’s organisational rules: interesting approach for modelling some

social aspects

12 Modelling the environment is important and could be done by

adopting abstractions from SODA

13 The concept of service as proposed in Gaia or PASSI should be

included in the methodology

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Extended requirements

5 Domains list (req. 1 and 3): problem, agency and solution domains 6 The layering principle (req. 3) coming from SODA will help in dealing

with complexity

7 Tropos experiences: goal-oriented analysis should be performed before

functional-oriented analysis

8 Interactions should include semantic communications 9 An ontology should be used to model agents’ knowledge 10 FIPA-compliance is advisable at least at the communication level 11 Gaia’s organisational rules: interesting approach for modelling some

social aspects

12 Modelling the environment is important and could be done by

adopting abstractions from SODA

13 The concept of service as proposed in Gaia or PASSI should be

included in the methodology

14 Non functional requirements should be explicitly modelled (req. 2) Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 7 / 21

Outline

1

The MEnSA Process Requirements

2

The New MEnSA Process

3

Results Assessment

4

Conclusions and Future Works

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 8 / 21

The MEnSA’s process composition

Our approach was inspired by PRoDe (PRocess for the Design of Design PRocesses) [Seidita et al., 2009]

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 9 / 21

The MEnSA’s process composition

Our approach was inspired by PRoDe (PRocess for the Design of Design PRocesses) [Seidita et al., 2009] We proposed some improvements during the fragments selection phase. . .

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 9 / 21

The MEnSA’s process composition

Our approach was inspired by PRoDe (PRocess for the Design of Design PRocesses) [Seidita et al., 2009] We proposed some improvements during the fragments selection phase. . . Our version combines the possibility of retrieving fragments directly on the basis

• f the

◮ process requirements ◮ metamodel as prescribed by PRoDe Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 9 / 21

The MEnSA’s process composition

The first step: collecting process requirements mainly from MEnSA project goals and team meetings

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 10 / 21

The MEnSA’s process composition

The first step: collecting process requirements mainly from MEnSA project goals and team meetings Fragments selection activity made available the set of fragments used to produce a first draft of the MAS metamodel

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 10 / 21

The MEnSA’s process composition

The first step: collecting process requirements mainly from MEnSA project goals and team meetings Fragments selection activity made available the set of fragments used to produce a first draft of the MAS metamodel

◮ fragments had been retrieved from the

repository according to the requirements they contribute to fulfill

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 10 / 21

The MEnSA’s process composition

The first step: collecting process requirements mainly from MEnSA project goals and team meetings Fragments selection activity made available the set of fragments used to produce a first draft of the MAS metamodel

◮ fragments had been retrieved from the

repository according to the requirements they contribute to fulfill

Once the metamodel had been polished, the initial set of fragments was positioned in a proper life-cycle

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 10 / 21

The MEnSA’s process composition

The first step: collecting process requirements mainly from MEnSA project goals and team meetings Fragments selection activity made available the set of fragments used to produce a first draft of the MAS metamodel

◮ fragments had been retrieved from the

repository according to the requirements they contribute to fulfill

Once the metamodel had been polished, the initial set of fragments was positioned in a proper life-cycle Therefore a proper process model had to be chosen

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 10 / 21

The MEnSA’s process composition

Fragment Assembly activity: the fragments had been positioned in the life-cycle place holders

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 11 / 21

The MEnSA’s process composition

Fragment Assembly activity: the fragments had been positioned in the life-cycle place holders Fragments Adaptation: solving incompatibility issues arising from the assembly of fragments coming from different processes

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 11 / 21

The MEnSA’s process composition

Fragment Assembly activity: the fragments had been positioned in the life-cycle place holders Fragments Adaptation: solving incompatibility issues arising from the assembly of fragments coming from different processes

◮ the fragments had been adapted to ⋆ properly support the new MAS

metamodel

⋆ comply with all input/output

constraints

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 11 / 21

The MEnSA’s process composition

Fragment Assembly activity: the fragments had been positioned in the life-cycle place holders Fragments Adaptation: solving incompatibility issues arising from the assembly of fragments coming from different processes

◮ the fragments had been adapted to ⋆ properly support the new MAS

metamodel

⋆ comply with all input/output

constraints

An initial version of the process had been available: this could be complete or not according to the refinements of the initial process requirements

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 11 / 21

The process component diagram

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

P Q R S T U V W X PQ PR PS PY PT PU

&'2)E*% F*G)-"*/'*+ 0"9#%1'2"3-4' :-+)(#4+ ,-".$'/% 0+#+'/'*+ >?'-#+)"* :E'*+

PP

!"#$ Z#-9;!"#$ >-E#*)M#+)"*#$

• 3$'

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 12 / 21

Outline

1

The MEnSA Process Requirements

2

The New MEnSA Process

3

Results Assessment

4

Conclusions and Future Works

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 13 / 21

Process evaluation

We have built a methodology that aims at filling the gap between existing AO methodologies and MAS infrastructures

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 14 / 21

Process evaluation

We have built a methodology that aims at filling the gap between existing AO methodologies and MAS infrastructures The sole creation of a new methodology by itself is not sufficient, also the MAS infrastructure needs to be re-conceived

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 14 / 21

Process evaluation

We have built a methodology that aims at filling the gap between existing AO methodologies and MAS infrastructures The sole creation of a new methodology by itself is not sufficient, also the MAS infrastructure needs to be re-conceived The methodology needs to be flexible and to accord its abstractions with the infrastructure’ ones

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 14 / 21

Process evaluation

We have built a methodology that aims at filling the gap between existing AO methodologies and MAS infrastructures The sole creation of a new methodology by itself is not sufficient, also the MAS infrastructure needs to be re-conceived The methodology needs to be flexible and to accord its abstractions with the infrastructure’ ones The work on the infrastructure is today going on, so at the moment is not possible to evaluate the all development process

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 14 / 21

Process evaluation

We have built a methodology that aims at filling the gap between existing AO methodologies and MAS infrastructures The sole creation of a new methodology by itself is not sufficient, also the MAS infrastructure needs to be re-conceived The methodology needs to be flexible and to accord its abstractions with the infrastructure’ ones The work on the infrastructure is today going on, so at the moment is not possible to evaluate the all development process Looking at the created methodology,

◮ it well satisfies the project requirements ◮ in the Design phase all the abstractions that are more

“infrastructure-like” (e.g. artifacts) have been adopted

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 14 / 21

Comparison regarding process-related criteria[Tran and Low, 2005]

MEnSA Gaia PASSI Tropos SODA Development lifecycle Iterative and incre- mental Iterative within each phase but sequential between phases Iterative across and within all phases (except for coding and deployment) Iterative and incre- mental Iterative and incre- mental Coverage of life- cycle Analysis and Design (and Implementa- tion) Analysis and De- sign Analysis, Design and Implementation Analysis and De- sign Analysis and De- sign Development perspective Middle-out Top-down Top- Down/Bottom-up (for pattern reuse) Top-down Middle-out Application domain Independent Independent Independent Independent Independent Size of MAS Not specified ≤ 100 agent classes Not specified Not specified Not specified Agent nature Heterogeneus Heterogeneus Heterogeneus BDI-like agents Heterogeneus Support for verification- validation Ongoing work No Yes Yes Ongoing work Ease

• f

under- standing of the process steps High High High High High Usability of the methodology Medium (guidelines not complete) Medium High Medium Medium Refinability Yes Yes Yes Yes Yes Approach to- wards MAS development

• a. i* framework and

OO b. RO (GO) a. OO b. RO (OrO)

• a. OO b. RO
• a. i* framework b.

NRO

• a. NOO b. RO

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 15 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project This is probably one of the largest SME experiments and probably its collaborative and distribution features make it unique

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project This is probably one of the largest SME experiments and probably its collaborative and distribution features make it unique It proved that several research groups can

◮ converge in defining a common project that aims at the definition of a

unique new methodology

◮ realise this project with a good degree of success by means of a novel

approach

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project This is probably one of the largest SME experiments and probably its collaborative and distribution features make it unique It proved that several research groups can

◮ converge in defining a common project that aims at the definition of a

unique new methodology

◮ realise this project with a good degree of success by means of a novel

approach

One of the major difficulties in MEnSA was the lack of automatic tools to compose the fragments

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project This is probably one of the largest SME experiments and probably its collaborative and distribution features make it unique It proved that several research groups can

◮ converge in defining a common project that aims at the definition of a

unique new methodology

◮ realise this project with a good degree of success by means of a novel

approach

One of the major difficulties in MEnSA was the lack of automatic tools to compose the fragments An higher-level tool which enables the composition of fragments and the production of new methodologies would be deserved

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Lessons learned

The new methodology is the result of the collaboration of four different research groups bringing their own experiences and expectations in the project This is probably one of the largest SME experiments and probably its collaborative and distribution features make it unique It proved that several research groups can

◮ converge in defining a common project that aims at the definition of a

unique new methodology

◮ realise this project with a good degree of success by means of a novel

approach

One of the major difficulties in MEnSA was the lack of automatic tools to compose the fragments An higher-level tool which enables the composition of fragments and the production of new methodologies would be deserved Such a “meta-tool” could also produce some “development tool” for the created methodology

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 16 / 21

Outline

1

The MEnSA Process Requirements

2

The New MEnSA Process

3

Results Assessment

4

Conclusions and Future Works

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 17 / 21

Conclusions

In this paper we have presented how we built a new methodology starting from the defined requirements and reusing fragments of existing methodologies

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 18 / 21

Conclusions

In this paper we have presented how we built a new methodology starting from the defined requirements and reusing fragments of existing methodologies Our approach for creating a new methodology started from PRoDe, but added some changes that permits to be more flexible in the process composition and in the fragments assembly

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 18 / 21

Conclusions

In this paper we have presented how we built a new methodology starting from the defined requirements and reusing fragments of existing methodologies Our approach for creating a new methodology started from PRoDe, but added some changes that permits to be more flexible in the process composition and in the fragments assembly In the paper we reported also how the created methodology meets the proposed requirements

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 18 / 21

Conclusions

In this paper we have presented how we built a new methodology starting from the defined requirements and reusing fragments of existing methodologies Our approach for creating a new methodology started from PRoDe, but added some changes that permits to be more flexible in the process composition and in the fragments assembly In the paper we reported also how the created methodology meets the proposed requirements We compared it with other methodologies, pointing out the advantages of the proposed new process in connection with the requirements

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 18 / 21

Future works

Even if we presented an almost completed methodology, a lot of work has still to be done for refining both the process and its metamodel

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 19 / 21

Future works

Even if we presented an almost completed methodology, a lot of work has still to be done for refining both the process and its metamodel

◮ defining of a standard process to assemble and adapt the different

fragments that now are analysed and assembled one-by-one in term of single inputs and outputs

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 19 / 21

Future works

Even if we presented an almost completed methodology, a lot of work has still to be done for refining both the process and its metamodel

◮ defining of a standard process to assemble and adapt the different

fragments that now are analysed and assembled one-by-one in term of single inputs and outputs

◮ completing our evaluation of the methodology Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 19 / 21

Future works

Even if we presented an almost completed methodology, a lot of work has still to be done for refining both the process and its metamodel

◮ defining of a standard process to assemble and adapt the different

fragments that now are analysed and assembled one-by-one in term of single inputs and outputs

◮ completing our evaluation of the methodology ◮ working on a supporting tool Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 19 / 21

Future works

Even if we presented an almost completed methodology, a lot of work has still to be done for refining both the process and its metamodel

◮ defining of a standard process to assemble and adapt the different

fragments that now are analysed and assembled one-by-one in term of single inputs and outputs

◮ completing our evaluation of the methodology ◮ working on a supporting tool ◮ continuing the test and evaluation of the methodology by using a case

study: the Bioinformatic Framework

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 19 / 21

Bibliography I

Cossentino, M., Gaglio, S., Garro, A., and Seidita, V. (2007). Method fragments for agent design methodologies: from standardisation to research. International Journal of Agent-Oriented Software Engineering, 1(1):91–121. Seidita, V., Cossentino, M., Hilaire, V., Gaud, N., Galland, S., Koukam, A., and Gaglio, S. (2009). The Metamodel: a Starting Point for Design Processes Construction. International Journal of Software Engineering and Knowledge Engineering. Tran, Q.-N. and Low, G. C. (2005). Comparison of ten agent-oriented methodologies. In Henderson-Sellers, B. and Giorgini, P., editors, Agent Oriented Methodologies, chapter XII, pages 341–367. Idea Group Publishing.

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 20 / 21

Building an Agent Methodology from Fragments: the MEnSA experience

Mariachiara Puviani & Massimo Cossentino Giacomo Cabri & Ambra Molesini

Dipartimento di Ingegneria dell’Informazione, Universit` a di Modena e Reggio Emilia, ICAR, Consiglio Nazionale delle Ricerche – Palermo, Alma Mater Studiorum—Universit` a di Bologna

AOMIP@SAC 2010, Sierre, Switzerland, 25th March 2010

Molesini (Univ. Bologna) MEnSA AOMIP 2010, 25/03/2010 21 / 21