Candidate interoperability standards: An ontological overlap analysis

Abstract

A web service choreography standard enables a standardized description of business processes that allows not only a clear specification of the control flow, but also forms the basis for the actual process execution. Such standards are part of the web services stack and facilitate enterprise systems interoperability. A simple indication of the rapid growth of this area is the number of new and existing standards. However, while the list of standards is expanding, there is an obvious lack of sound theoretical evaluation and comparison of these standards. To provide a basis for this theoretical evaluation and comparison, we use the well-established Bunge–Wand–Weber (BWW) models, in particular the representation model. We, and other researchers, have shown the usefulness of the BWW representation model for analyzing, evaluating, and engineering techniques in the areas of traditional and structured systems analysis, object-oriented modeling, and process modeling. In this paper, we present the BWW ontological analysis of four dominating candidate interoperability standards; ebXML Business Process Specification Schema BPSS (v1.01), the Business Process Modeling Language BPML (v1.0), the Business Process Execution Language for Web Services BPEL4WS (v1.1) and the Web Service Choreography Interface WSCI (v.1.0). The application of the BWW representation model and the overlap analysis across the four standards has resulted in several interesting findings. First, we have identified a set of ontological constructs for which representations are missing in each of the standards. This situation has allowed us to generate a number of propositions as to the potential weaknesses of the standards in general. Second, the analysis has shown that ebXML BPSS is a “clear winner” with respect to the ontological expressiveness and the maturity of the specification. Furthermore, the analysis has also allowed us to identify two sets of standards viz. ebXML BPSS and BPEL4WS, ebXML BPSS and WSCI, that, when used together, provide the user with the largest available set of meaningful constructs required to implement systems interoperability. Finally, the analysis explains which standards are complementary to each other and which standards are in fact competing standards.

Introduction

Enterprise systems and other software packages, together with middleware solutions such as workflow management and enterprise application integration, have facilitated the integration of functions and data within many organizations. After more than a decade of such integration initiatives, an extension of the scope of integration management is apparent. The integration of business processes between organizations becomes now a major focus. The interoperability of enterprise systems or the integration of cross-organizational business processes in general, requires a clear specification of the information flow between the participating business partners. The extensible mark-up language, XML, provided a new approach to describe such standardized documents and systems. It also allowed the development of autonomous and modular solutions that can be published, identified, and accessed independently from the technological platform. This set of solutions is commonly termed web services.

Various XML-based standards have been developed and proposed for the area of enterprise systems interoperability (ESI) and web services. These standards cover different layers of the web services stack [25] and range from very technical specifications to languages for executable business processes. However, underlying all of these candidate standards is the fact that they are attempting to emulate real-world entities as well as the ways they interact to perform such useful tasks as business processes (sales, purchasing, and fulfillment) through computer systems interoperation. Accordingly, all of these standards are based on some model of how the real-world entities and processes are expected to operate.

Most of these standards have not been evaluated or even compared on a theoretical foundation. Software vendors, internal IT departments, external service providers, and IT project managers, as well as the developers of these standards, however, require guidance in the development, selection, combination and application of these standards. Previous analyses are based on logical reasoning and analytical arguments. On this base it is difficult to strive for completeness or objectivity. A theoretical evaluation, on the other hand, is grounded in a well-developed, generally accepted model, which could be, for example, a foundational ontology.

This situation can be compared with the traditional area of Information Systems Analysis and Design (ISAD) techniques. Many modeling techniques have been developed for the description of data, functions, objects, processes, and the like. The development, evaluation, and selection of these techniques, however, have been difficult due to the lack of a generally accepted theoretical foundation that could serve as a benchmark.

One promising benchmark for the analysis of ISAD grammars has been the collection of ontological models developed by Bunge, Wand and Weber [45], [46], [47], [48]. Based on sound philosophical foundations, these models provide a well-defined list of constructs and interrelationships that are perceived as relevant and which can be used as a platform for the analysis of such grammars. The extent to which a grammar does not cover all BWW constructs and the extent to which the elements of a grammar go beyond the BWW models provide valuable insights into the potential shortcomings of that grammar. Many of the most popular modeling techniques have now been discussed in the light of their comparability with the BWW models (see Table 2). While the majority of previous ontological analyses has been applied to conceptual modeling techniques, they can also be applied to models which are closer to technical specifications such as interoperability standards.

Currently, four candidate standards are perceived as dominating discussion in the web services area, viz. ebXML Business Process Specification Schema (ebXML BPSS, v1.01), the Business Process Modeling Language (BPML, v1.0), the Business Process Execution Language for Web Services (BPEL4WS, v1.1) and the Web Service Choreography Interface (WSCI, v1.0) (see, for example, [2]). The aim of this research is to analyze and evaluate each of those four standards in order to identify their potential weaknesses as well as to identify the set of standards which, when used in combination, provides users with the largest available set of meaningful constructs required to implement systems interoperability. This research is motivated in several ways. First, in a field where a number of competing standards are being proposed, we want to provide some theoretically based guidance to practitioners on the strengths and weaknesses of these standards. Second, in a similar manner, we want to provide a theoretically based evaluation that can demonstrate some potential areas for further refinement to the standard developers. Third, given that each standard may display deficiencies, we wish to investigate what combinations of the analyzed standards might be suggested to provide users with the largest available set of meaningful constructs required to implement systems interoperability. The success of BWW analysis of ISAD grammars and the existing analogy between the ISAD grammars and ESI standards, because both are attempting to model real-world interactions (e.g. buying, selling, paying), implies that such analysis would be fruitful.

The study confirms that there are several weaknesses which exist across all four candidates and develops a set of propositions for future empirical testing in order to understand the relevance of the weaknesses for the ESI domain. The study also identifies ebXML BPSS (v1.01) as being the most ‘mature’ standard and further indicates two pairs of standards, viz. ebXML BPSS and BPEL4WS (v1.1), ebXML BPSS and WSCI (v1.0), that should be used in combination to provide users with the largest available (but minimally ontologically overlapping) set of meaningful constructs required to implement systems interoperability. Moreover, the analysis also explains the dichotomy of complementary and competing standards.

This paper is structured as follows. Section 2 introduces the selected theoretical foundation for this research, i.e. the Bunge–Wand–Weber models, in particular the representation model. Section 3 reviews the related research. While this review presents prior work involving the BWW models generally, it focuses on the research related to ESI standards. Section 4 outlines the research methodology, which is an extension of the ‘classical’ ontological analysis. The actual outcomes of this research and the resulting propositions are presented in Sections 5 BWW representation mapping analysis, 6 Overlap analysis and propositions. Section 5 provides a detailed analysis of BPEL4WS (v1.1) as an example to show the capabilities of an ontological analysis on an individual candidate standard. Section 6 demonstrates the results of an ontological overlap analysis across the four selected candidate standards. The paper concludes with a discussion of limitations and future work required in this area.