Semantic Annotation for Process Models: Facilitating Process Knowledge Management via Semantic Interoperability

Business process models representing process knowledge about doing business are necessary for designing Information Systems (IS) solutions in enterprises. Interoperability of business process knowledge in legacy systems is crucial for enterprise systems interoperation and integration due to increased enterprise cooperation and business exchange. Many modern technologies and approaches are deployed to support business process interoperability either at the instance level or the protocol level, such as BPML, WSDL and SOAP. However, we argue that a holistic approach is necessary for semantic interoperability of business process models at the conceptual level when considering the process models as reusable process knowledge for other (new or integrated) IS solutions. This brings requirements to manage semantic heterogeneity of process knowledge in process models which are distributed across different enterprise systems. Semantic annotation is an approach to achieve semantic interoperability of heterogeneous resources. However, such an approach has usually been applied to enhance the semantics of unstructured and structured artifacts (e.g. textual resources [72] [49], and Web services [166] [201]). The aim of the research is to introduce an ontology-based semantic annotation approach to enrich and reconcile semantics of process models — a kind of semi-structured artifact, for managing process knowledge. The approach brings together techniques in process modeling, ontology building, semantic matching, and Description Logic inference in order to provide a comprehensive semantic annotation framework. Furthermore, a prototype system that supports the process of ontology-based semantic annotation of heterogeneous process models is described. The applicational goal of our approach is to facilitate process knowledge management activities (e.g. discovery, reuse, and integration of process knowledge/models) by enhanced semantic interoperability. A survey has been performed through identifying semantic heterogeneity in process modeling and investigating semantic technology from theoretical and practical views. Based on the results from the survey, a comprehensive semantic annotation framework has been developed, which provides a method to manage semantic heterogeneity of process models from the following perspectives. First, basic descriptions of process models (profile annotation); second, process modeling languages (meta-model annotation); third, contents of process models (model annotation) and finally intentions of process model owners (goal annotation). Applying the semantic annotation framework, an ontology-based annotation method has been elaborated, which results in two categories of research activity — ontology building and semantic mapping. In ontology building, we use Web Ontology Language (OWL), a Semantic Web technology, which can be used to model ontologies. GPO (General Process Ontology) comprising core concepts in most process modeling languages is proposed; domain concepts are classified in the corresponding categories of GPO as a domain ontology; design principles for building a goal ontology are introduced in order to serve the annotation of process models pragmatically. In semantic mapping, a set of mapping strategies are developed to conduct the annotation by considering the semantic relationships between model artifacts and ontology references and as well the semantic inference mechanism supported by OWL DL (Description Logic). The annotation method is finally formalized into a process semantic annotation model - PSAM. The proposed approach has been implemented in a prototype annotation tool —ProSEAT to facilitate the annotation process. Procedures of applying the semantic annotation approach with the tool are described through exemplar study. The annotation approach and the prototype tool are evaluated using a quality framework. Furthermore, the applicability of the annotation results is validated by going through a process knowledge management application. The Semantic Web Rule Language (SWRL) is applied in the application demonstration. We argue that the ontology-based annotation approach combined with the Semantic Web technology is a feasible approach to reconcile semantic heterogeneity in the process knowledge management. Limitations and future work are discussed after concluding this research work. The contributions of this thesis are summarized as follows. First, a general process ontology is proposed for unifying process representations at a high level of abstraction. Second, a semantic annotation framework is introduced to describe process knowledge systematically. Third, ontology-based annotation methods are elaborated and formalized. Fourth, an annotation system, utilizing the developed formal methods, is designed and implemented. Fifth, a process knowledge management system is outlined as the platform for manipulating the annotation results. Moreover, applying results of the approach is demonstrated through a process model integration example