Page 146 -
P. 146

AGENT-ORIENTED  METHODS  AND  METHOD  ENGINEERING     131

                    Table 8.1

                    Comparison of Ten Agent-Oriented Methodologies (continued)
                    D. Comparison Regarding Supportive-Related Criteria
                                                                MAS-CommonKADS






                                                       Gaia  Tropos  Prometheus  PASSI  Adelfe  MaSE  RAP  MESSAGE  INGENIAS

                    Software and methodological support  N  N  N    Y   Y    Y   Y    Y    Y   Y
                    Open systems                      Y   N    N    N   N    Y   N    N    N   N
                    Dynamic structure                 N   N    N    N   Y    N   P    N    N   N
                    Agility and robustness            N   N    Y    Y   N    Y   Y    Y    N   N
                    Support for conventional objects  N   N    N    Y   N    Y   N    Y    Y   Y
                    Support for mobile agents         N   N    N    N   Y    N   N    N    N   N
                    Support for ontology              N   N    Y    N   Y    N   N    N    Y   N
                      Source: After Tran and Low (2005).
                      Key: H = high; M = medium; L = low; Y = yes; N = no; P = possibly.


                    a high degree of flexibility to the process engineer, perhaps assisted by an automated tool (Nguyen
                    and Henderson-Sellers, 2003), who can allocate appropriate deontic values to any specific pair
                    of process components depending upon the context, that is, the specific project, skills set of the
                    development team, and so on. Linkage decisions are made either subjectively/experientially or by
                    means of an overall assessment of a number of factors relating to the project. These factors include
                    maturity/capability level (such as CMM or SPICE), specific skills in the workforce, domain of
                    the project, and so forth.
                      As noted earlier, one of the hardest tasks currently in SME construction is the selection of the
                    optimal set of method fragments to suit any specific situation. Syntactic coupling can be verified in
                    terms of the matching of the output from one fragment to the input for a second. This is facilitated
                    by generating fragments from a metamodel and also by using a standard way of documenting the
                    fragments. Nevertheless, the current reality is that the semantic aspect of the fragments must be
                    analyzed “by hand,” usually by a skilled method engineer (either in-house or as a visiting con-
                    sultant or mentor). Work toward a more objective approach is under way (e.g., McBride, 2004;
                    Nguyen and Henderson-Sellers, 2003; Ralyté, 2004), and prototype tools (MethodComposer,
                    MET) have been developed.
                      Creation of a project-specific or organization-specific agent-oriented methodology then proceeds
                    using the agent-oriented method fragments contained in the OPF repository together with many
                    non–agent-oriented method fragments (typically the earlier object-oriented method fragments) that
                    are needed for those elements of software development that are not technology/paradigm-dependent.
                    These include method fragments to describe project management, some metrics, reusability, and
                    so on. A fully comprehensive methodology, suitable for direct industry usage, can be constructed
   141   142   143   144   145   146   147   148   149   150   151