Article Preview
TopIntroduction
Increasing numbers of enterprises are choosing to integrate their business processes with each other in order to facilitate collaboration between them, e.g., through workflow integration and Web Services orchestration. The use of models of integrated business processes can provide a common platform for process integration and Web Services orchestration. The process models of enterprises that are currently in use (legacy models) can be treated as reusable assets, from which knowledge of the processes used by those enterprises can be extracted, thereby facilitating the building of integrated process models. Knowledge of these processes provides a guide for finding common ground between the processes used by the different organizations, and thus provides a framework for optimizing the integration of the existing processes. However, there is significant heterogeneity in the process models that are used in individual enterprises: often, diverse business jargon and different modeling languages are used. This heterogeneity can lead to problems with respect to the extraction of process knowledge and its use in building an integrated process model. The reconciliation of semantic heterogeneity has been identified as a very laborious task in integration projects (Doan, Noy, & Halevy, 2004). In this paper, we focus on the semantic interoperability of business process models.
Ontology-based semantic annotation is generally considered to be an appropriate technique for achieving semantic interoperability, and is achieved by introducing common means of understanding and standardization. Most semantic annotation work has been developed on and applied to unstructured (e.g., MnM (Vargas-Vera, Motta, Domingue, Lanzoni, Stutt, & Ciravegna, 2002), KIM (Popov, Kiryakov, Kirilov, Manov, Ognyanoff, & Goranov, 2004), AeroDAML (Kogut & Holmes, 2001) and OntoMat-Annotizer (Handschuh, Staab, & Maedche, 2001) for textual resources) and structured (e.g., METEROR-S (Patil, Oundhakar, Sheth, & Verma, 2004), WSMO (Bruijn, Bussler, Domingue, Fensel et al., 2005), OWL-S (Martin, Burstein, Hobbs, Lassila, McDermott, Mcllraith et al., 2004), SAWSDL (Kopecky, Vitvar, Bourne, & Farrell, 2007) and WSDL-S (Akkiraju, Farrell, Miller, Nagarajan et al., 2005) for Web services) artifacts to improve interoperability at different levels. A few semantic approaches (including our own) that use semi-structured artifacts (usually enterprise models) have been developed in recent years, such as those reported in INTEROP (Interoperability Research for Networked Enterprises Applications and Software; Panetto, Scannapieco, & Zelm, 2004) and ATHENA (Advanced Technologies for Interoperability of Heterogeneous Enterprise Networks and their Applications; Ruggaber, 2006). For the semantic annotation of unstructured and structured artifacts, semantic reconciliation focuses on one level only, that of the data in the text or in the schema. For semi-structured artifacts, semantic heterogeneities are taken into account on more than one level. These are usually either the meta-model (i.e., modeling language) or model levels (i.e., model content), or even the intention level (i.e., goal modeling).