Agent-Oriented Software Engineering PORTO Methodology AIAD 2013/2014. António Castro and Eugénio Oliveira

Transcription

1 Agent-Oriented Software Engineering PORTO Methodology AIAD 2013/2014 António Castro and Eugénio Oliveira NIAD&R Distributed Artificial Intelligence and Robotics Group 1 Contents What is AOSE? Main Existing Methodologies Tropos AUML Gaia Porto NIAD&R Distributed Artificial Intelligence and Robotics Group 2 1

2 What is AOSE? A methodological approach for the development of software oriented or based on agents (Giorgini and Henderson-Sellers) A methodology has two main components: A Description of approach processes. Documentation and products/artifacts. All existing proposals are from the academic world, are not complete (either in processes or artifacts/products) nor applicable to a wider scope of application domains. NIAD&R Distributed Artificial Intelligence and Robotics Group 3 Main existing methodologies Name Characteristics Authors TROPOS Analysis/Design; Supports Verification/validation Paolo Giorgini et al. MAS- CommonKADS PASSI PROMETHEUS GAIA ADELFE MESSAGE INGENIAS Analysis/Design; Guidelines for Verification/Validation Analysis/Design/Implementation; Supports Verification/validation; Analysis/Design; Supports Verification/validation; QQ # agts. Analysis/Design; Does not Support Verification/validation; <= 100 agts. Analysis/Design/Implementation; Supports Verification/validation; Analysis/Design; Does not Support Verification/validation; Analysis/Design/Implementation; Supports Verification/validation; Carlos Iglesias et al. Massimo Cossentino Lin Padgham et al. Franco Zambonelli et al. Carole Bernon et al. Francisco Garijo et al. Juan Pavón et al. PORTO Analysis/Design/Implementation/Testing/Validation Antonio Castro et al. NIAD&R Distributed Artificial Intelligence and Robotics Group 4 2

3 TROPOS P. Bresciani, P. Giorgini, F. Giunchiglia, J. Mylopoulos and A. Perini. TROPOS: An Agent-Oriented Software Development Methodology. In Journal of Autonomous Agents and Multi-Agent Systems, Kluwer Academic Publishers Volume 8, Issue 3, Pages , May 2004 NIAD&R Distributed Artificial Intelligence and Robotics Group 5 Main Characteristics BDI specific Develops the system using BDI constructs Emphasis on early requirement analysis NIAD&R Distributed Artificial Intelligence and Robotics Group 6 3

4 Early Requirements Analysis Model needs and objectives of various system stakeholders in the human organization Model the system environment Model interaction of system-to-be with human stakeholders and system environment NIAD&R Distributed Artificial Intelligence and Robotics Group 7 Example Actor Diagram NIAD&R Distributed Artificial Intelligence and Robotics Group 8 4

5 Example Goal Diagram NIAD&R Distributed Artificial Intelligence and Robotics Group 9 Main Concepts Use actor to represent agents, roles and positions Use goals, plans, resources, dependencies, capabilities and beliefs throughout development NIAD&R Distributed Artificial Intelligence and Robotics Group 10 5

6 Strengths and Weaknesses Strength Good support for early requirement analysis Weakness BDI specific, less general Diagrams may not scale NIAD&R Distributed Artificial Intelligence and Robotics Group 11 AUML J. Odell, H. Van Dyke Parunak and B. Bauer Representing Agent Interaction Protocols in UML in Agent-Oriented Software Engineering, Paolo Ciancarini and Michael Wooldridge eds., Springer-Verlag, Berlin, pp , NIAD&R Distributed Artificial Intelligence and Robotics Group 12 6

7 AUML Not a methodology Standards to represent design of agent systems Familiar to industry developers Widely adopted NIAD&R Distributed Artificial Intelligence and Robotics Group 13 GAIA F. Zambonelli, N. Jennings, M. Wooldridge Developing Multiagent Systems: the Gaia Methodology ACM Transactions on Software Engineering and Methodology, Vol. 12, No. 3, July 2003 NIAD&R Distributed Artificial Intelligence and Robotics Group 14 7

8 Overview The goals of the organizations that constitute the overall system and their expected global behavior The rules that the organization should respect and enforce in its global behavior OUTPUT Separating, when possible, the organizational independent aspects (detected in analysis) from the organizational dependent ones (derived from the adoption of a specific organizational structure). OUTPUT OUTPUT OUTPUT In terms of its topology and control regime. Can also exploit catalogues of organizational patterns. Identifies the agent classes that will make up the system and the agent instances that will be instantiated from these classes. COMPLETION DEFINITION COMPLETION DEFINITION Identifies the main services coherent blocks of activity in which agents will engage that are required to realize the agent s roles and properties. NIAD&R Distributed Artificial Intelligence and Robotics Group 15 Analysis Phase Organizations: determine whether multiple organizations have to coexist in the system and become autonomous interacting MAS. Environmental Model: abstract computational resources, such as variables or tuples, made available to the agents for sensing (read), for effecting (change) or for consuming (extract) NIAD&R Distributed Artificial Intelligence and Robotics Group 16 8

9 Analysis Phase Preliminary Role Model: identify the basic skills that are required by the organization to achieve its goals, as well as the basic interactions that are required for the exploitation of these skills. Preliminary Interaction Model: captures the dependencies and relationships between the various roles in the MAS organization, in terms of one protocol definition for each type of inter role interaction. NIAD&R Distributed Artificial Intelligence and Robotics Group 17 Analysis Phase Organizational Rules: responsibilities of the organization as a whole. These are safety (invariants that must be respected) and liveness (dynamics of the organization) organizational rules. Preliminary Interactions Preliminary Roles Example of an Analysis Diagram Environment Model NIAD&R Distributed Artificial Intelligence and Robotics Group 18 9

10 Architectural Design Including a detailed description of the semantics of the relations Organizational Structure: identify the appropriate organizational structure, including, topology and control regime. Organizational Patterns: catalogue of possible modular and composable organizational structures that will help the designer. Graphical Representation Formal Representation NIAD&R Distributed Artificial Intelligence and Robotics Group 19 Architectural Design Completion of Role and Interaction Models: (1) define all the activities in which a role will be involved, (2) define organizational roles, (3) complete the definition of the protocols required by the application and (4) define organizational protocols. Role Model Interaction Model NIAD&R Distributed Artificial Intelligence and Robotics Group 20 10

11 Detailed Design Agent Model: to define the agent model it is necessary to identify which agent classes are to be defined to play specific roles and how many instances of each class have to be instantiated in the actual system. The model can be defined using a simple diagram (or table) specifying, for each class, which roles will map it. In addition, the model can document the instances of a class that will appear in the MAS. Agent Class Role NIAD&R Distributed Artificial Intelligence and Robotics Group 21 Detailed Design Services Model: identify the services associated with each agent class, or equivalently, with each of the roles to be played by the agent classes. For each service it is necessary to document its properties: inputs, outputs, preconditions and postconditions. The services are derived from the list of protocols, activities, responsibilities and liveness properties of the roles it implements. NIAD&R Distributed Artificial Intelligence and Robotics Group 22 11

12 Scopes and Limitations Does not directly deal with particular modeling techniques. It proposes but does not commit to, specific techniques for modeling (e.g., roles, environment, interactions). In the future: AUML is a useful companion to GAIA. Does not directly deal with implementation issues. The outcome is a detailed but technology-neutral specification. Should be easy to implement with, for example, a FIPA-compliant agent system. Does not explicitly deal with activities of requirements capturing and modeling. In the future: integrate methods and techniques from goaloriented analysis. NIAD&R Distributed Artificial Intelligence and Robotics Group 23 Summary of Strengths Expressive except does not model knowledge in system Architecturally independent, allow diverse technologies Simple and easy to learn NIAD&R Distributed Artificial Intelligence and Robotics Group 24 12

13 Summary of Weaknesses Does not cover full life-cycle Unfamiliar notation for developers Lack of CASE tool and other support NIAD&R Distributed Artificial Intelligence and Robotics Group 25 PORTO better than GAIA Antonio Castro. "A Distributed Approach to Integrated and Dynamic Disruption Management". PhD Thesis, FEUP, University of Porto, Portugal. July Antonio Castro & Eugenio Oliveira. "The Rationale behind the Development of an Airline Operations Control Centre using GAIA based Methodology". International Journal of Agent-Oriented Software Engineering, Vol. 2, No. 3, pp NIAD&R Distributed Artificial Intelligence and Robotics Group 26 13

14 PORTO Overview/1 NIAD&R Distributed Artificial Intelligence and Robotics Group 27 PORTO Overview/2 NIAD&R Distributed Artificial Intelligence and Robotics Group 28 14

15 PORTO Overview/3 NIAD&R Distributed Artificial Intelligence and Robotics Group 29 Requirements Analysis NIAD&R Distributed Artificial Intelligence and Robotics Group 30 15

16 Advantages Modelling specifications in terms of actors, their roles, their goals and dependencies, is more similar to the AOCC organisation. In subdividing the system: identifying the specific organisations and sub-organisations dedicated to the achievement of a specific goal. In preliminary role model: identifying the basic skills (functionalities and competences) required by the organisation to achieve its goals. In preliminary interaction model: identifying the basic interactions that are required for the exploitation of the basic skills. NIAD&R Distributed Artificial Intelligence and Robotics Group 31 Analysis Preliminary Role & Interaction NIAD&R Distributed Artificial Intelligence and Robotics Group 32 16

17 Analysis - Combined Representation NIAD&R Distributed Artificial Intelligence and Robotics Group 33 Arch. Design Org. Rules and Structure NIAD&R Distributed Artificial Intelligence and Robotics Group 34 17

18 Organisational Rules and Structure NIAD&R Distributed Artificial Intelligence and Robotics Group 35 Organisational Rules and Structure NIAD&R Distributed Artificial Intelligence and Robotics Group 36 18

19 Arch. Design - Combined Representation NIAD&R Distributed Artificial Intelligence and Robotics Group 37 Arch. Design - Completing role model NIAD&R Distributed Artificial Intelligence and Robotics Group 38 19

20 Completing interaction model NIAD&R Distributed Artificial Intelligence and Robotics Group 39 Final Role, Interaction and Environment NIAD&R Distributed Artificial Intelligence and Robotics Group 40 20

21 Detail Design - Agent Model NIAD&R Distributed Artificial Intelligence and Robotics Group 41 Detail Design Agent Model NIAD&R Distributed Artificial Intelligence and Robotics Group 42 21

22 Detail Design - Service Model NIAD&R Distributed Artificial Intelligence and Robotics Group 43 Detail Design Service Model NIAD&R Distributed Artificial Intelligence and Robotics Group 44 22

23 Implementation Id. Concepts & Actions NIAD&R Distributed Artificial Intelligence and Robotics Group 45 Implementation - Mapping NIAD&R Distributed Artificial Intelligence and Robotics Group 46 23

24 Implementation - Development NIAD&R Distributed Artificial Intelligence and Robotics Group 47 Testing and Validation Test Cases NIAD&R Distributed Artificial Intelligence and Robotics Group 48 24

25 Thanks for your attention! Personal Web Pages MASDIMA Project NIAD&R Distributed Artificial Intelligence and Robotics Group 49 25