SEMANTIC INTEROPERABILITY AND PROCESS ONTOLOGIES 45 linking the parameters to a resource class which is defined separately from the process

ontology.

The Intelligent Software Agents Lab at Carnegie Mellon University has played a critical role in the development of OWL-S from its very conception. In addition, the Softagent Group has also developed the most complete and integrated set of OWL-S development tools, the OWL-S IDE. However, OWL-S is criticized to suffer conceptual ambiguity, lack concise axiomatization, be designed too loosely and offer an overly narrow view on Web services [106].

3.2.7 WSMO (Web Service Modeling Ontology)

WSMO is a conceptual model for describing various aspects related to Semantic Web services. The objective of WSMO and its accompanying efforts is to solve the ap-plication integration problem for Web services by defining a coherent technology for Semantic Web services [210].

WSMO provides ontological specifications for the core elements of Semantic Web services. The representation of WSMO follows the MOF (Meta-Object Facility) [124]

specification, and the MOF constructs Class, sub-Class, Attributes and type are used in the definition of WSMO. Every construct of WSMO is called a WSMO ele-ment. There are five top-level elements: annotations, ontologies, Web services, goals and mediators. WSMO does not contain the constructs to represent any of the pro-cess perspectives, but only provide interface to describe the functionality of the Web service. A twofold view on the operational competence of the Web service is provided:

choreography decomposes a capability in terms of interaction within the Web service;

orchestration decomposes a capability in terms of functionality required form other Web services. The process of Web services composition is excluded from WSMO. Another important feature of WSMO is that it introduces the elements Goal and Mediator.

Goal can represent an objective of the execution of a Web service. Mediator is exploited to overcome interoperability problems between different WSMO elements. For example, a wgMediator (Web service to goal Mediator) links Web services to goals, indicating that the Web service (totally or partially) fulfills the goal to which it is linked. Such a mediator may explicitly state the difference between the two entities and map different vocabularies through the use of ooMediators (ontology to ontology Mediators) [209].

3.2.8 POP* (Process, Organization, Product and others)

POP* methodology is one of the research contributions of the EU project ATHENA [139]. The overall objective of ATHENA is to enable enterprise to seamlessly inter-operate with others. The POP* methodology aims to develop a set of core modeling methodology elements for capturing collaborative enterprises design and management.

It offers a model exchange device by providing a common format along with a mapping methodology to define the mappings from the various enterprise modeling languages to the common format. The POP* methodology includes a common format as the ex-change format — the POP* meta-model containing a set of basic modeling constructs.

There are five dimensions included in POP*, namely the Process, Organization, Prod-uct, Decision and Infrastructure dimensions [138].

46 CHAPTER 3. STATE OF THE ART

Figure 3.8: The POP* meta model for Process dimension [138]

The POP* meta-model is represented in UML class diagram to specify the concepts and their interrelations (including the cardinality). The properties of each concept and relationship are defined by property names and value types. Therefore, POP* can be described in the UML 2.0 Profile as a basis for further implementation of POP* as an enterprise modeling language [138]. Figure 3.8 provides an overview of the modeling constructs in the Process dimension.

Process is the central construct and the other constructs associated with it can well support the representations of the operational/functional, structural, control and re-sources perspectives. Transactions of state and data are implicitly represented through the constructs event and process role carried by flow. Th organizational perspective is not included in the Process dimension in this version of POP*. Some organizational concepts are introduced in the Organizational dimension, but no specific relationships are defined to associate them to process.

3.2.9 UEML2 (Unified Enterprise Modeling Language version 2) UEML was initiated by the UEML project [130] to provide industry with a unified and expandable modeling language. UEML 2.0 and UEML 2.1 are further developed in the INTEROP-NoE project [140]. The work on UEML2 in this project is aimed on char-acterising correspondences between enterprise modeling languages to enable model interchange. A "UEML template approach" is applied in the development of UEML2.1.

This approach requires a detailed (ontological) analysis of the constructs found in en-terprise modeling languages and allows to formally define correspondences between constructs in distinct languages and thereby a UEML-based core enterprise modeling

3.2.SEMANTICINTEROPERABILITYANDPROCESSONTOLOGIES47

Table 3.2: Ontological representations of different process ontologies

BWW on-tology

MIT pro-cess hand-book

TOVE PSL PIF OWL-S POP* UEML

Representation primitives concept set

- status state - result - extension

of changin- gAttribut-edThing

48 CHAPTER 3. STATE OF THE ART language [141]. The meta-model level of UEML is based on the BWW ontology but the UEML approach encourages the addition of new ontological classes, properties, states and transformations when describing modeling constructs. The growth of the BWW ontology is referred to as the UEML ontology [142]. Figure 3.9 displays the main UEML ontology classes. Due to the extensibility of the core ontology, the UEML ontology is able to cover all the process perspectives through the proper extensions.

Figure 3.9: Generalization hierarchy of UEML ontology classes [144]

The UEML ontology are represented using OWL Full with the Protégé as the UEMLBase tool. The UEMLBase tool has been used for experimentally describing the following languages: (Class and Activity diagrams of) UML 2.0, CPN (Coloured Petri Net), GRL (Goal-oriented Requirements Language), KAOS (Knowledge Acqui-sition in autOmated Specification), IDEF3 (Integrated DEFinition methods – Process Description Capture), and, partially, ISO 19440; the result is the UEMLBase content that can be retrieved, visualized and navigated by using the UEMLBase tool function-ality [142].

3.2.10 Comparison of process ontology representations

Table 3.2 displays the comparison of ontology representation primitives, and main con-cepts and representation mechanisms for the process perspectives. The comparison presents how the semantics of process, i.e. the process perspectives are specified in dif-ferent ontological representation forms. Some of them are specified like a meta-model of process modeling languages, such as PIF, OWL-S, POP*. Some are very general and need to be extended to represent more specific process semantics, such as BWW and UEML. The MIT process handbook and TOVE provide process templates and stan-dardized taxonomy. WSMO is unique from the other ontologies by the use of interfaces to describe the functionality of Web services, so that there are no particular ontological representations for the process perspectives but mainly for the mediation of services.