Page 135 -
P. 135

120     HENDERSON-SELLERS
                    Figure 8.1  Genealogy of Various Agent-Oriented Methodologies and Their Relationships to
                             Object-Oriented Methodologies
                          i*
                                    Tropos
                                             MAS-CommonKADS                 Agent Factory
                           CAMLE                 (+AI/KE)

                                                                       Cassiopeia
                                                           MaSE
                           INGENIAS
                                                                                         ROADMAP
                                                                               SODA
                                                                 Kendall
                                                                 et al.
                         MESSAGE        Adelfe
                       RAP                              AAII             Gaia          Agent OPEN




                                  RUP/UP          OMT               Fusion        OPEN
                    AOR



                                                          OO
                                     PASSI
                                                                           Prometheus

                      Source: Modified from Henderson-Sellers (2005).

                      There are several AO methods acknowledging an OO influence, either from OMG, RUP, or
                    OPEN (the last of these is discussed in more detail below).
                      MaSE (DeLoach, 1999; Wood and DeLoach, 2000) is drawn from the legacy of object-oriented
                    methodologies such as OMT together with influences from the more recent UML as well as pre-
                    existing work in the realm of agents and multi-agent systems (e.g., Kinny, Georgeff, and Rao,
                    1996; Kendall and Zhao, 1998). It aims to guide the designer through the multi-agent system
                    development process from an initial system specification to a set of formal design documents. It
                    has two phases: analysis and design. The former deals with the specification of system goals, use
                    cases, sequence diagrams, roles, and tasks, while the latter uses the analysis phase’s outputs to
                    design agent classes, agent interactions, and agents’ internal components. It is also well supported
                    by a software tool. Another input to MaSE (Figure 8.1) is the work of Zhang, Kendall, and Jiang
                    (2002) (referred to hereafter as the ZKJ methodology). The ZKJ methodology focuses on the iden-
                    tification of goals and roles. It represents the process by a set of ten “activities,” each having an
                    input, an output, a control, and a mechanism. Four of these activities (Identify actors, Identify use
                    cases, Identify objects, and Determine business objects) are grouped as “object-oriented analysis
                    activities,” the rest being focused on agent goals and roles. There are six activities focused on
                    roles and goals (Identify goals, Develop goal cases, and Identify beliefs, Identify roles, Assign
                    goals to responsibilities, Assign and compose roles, and Identify composite roles). Indeed, role
                    identification is an important activity in the ZKJ methodology. Roles are seen as able to execute
                    a set of activities in order to fulfill one or more responsibilities.
   130   131   132   133   134   135   136   137   138   139   140