No abstract

Business system designers want to integrate heterogeneous legacy systems to provide flexible business services cheaper and faster. Unfortunately, modern integration technologies represent important integration knowledge only implicitly making solutions harder to understand, verify, and maintain. In this paper we propose a data-driven approach, "Semantically-Enabled Externalization of Knowledge" (SEEK), that explicitly models the semantics of integration requirements & capabilities, and data transformations between heterogeneous legacy systems. Goal of SEEK is to make the systems integration process more efficient by providing tool support for quality assurance (QA) steps and generation of system configurations. Based on use cases from industry partners, we compare the SEEK approach with UML-based modeling. In the evaluation context SEEK was found to be more effective to make expert knowledge on system requirements and capabilities available for more efficient tool support and reuse.

The capability to provide a platform for flexible business services in the Air Traffic Management (ATM) domain is both a major success factor for the ATM industry and a challenge to integrate a large number of complex and heterogeneous information systems. Most of the system knowledge needed for integration is not available explicitly in machineunderstandable form, resulting in time-consuming and error-prone human tasks.In this paper we propose a knowledge-based approach, "Semantically-Enabled Externalization of Knowledge" for the ATM domain (SEEK-ATM), which explicitly models a) expert knowledge on specific heterogeneous systems and integration requirements; and b) allows mapping of the specific knowledge to the general ATM problem domain knowledge for semantic integration. The domain-specific modeling enables a) to verify the integration knowledge base as requirements specification for later design of technical systems integration and b) to provide an API to the problem space knowledge to facilitate tool support for efficient and effective systems integration.Based on an industry case study, we evaluate effects of the proposed SEEK-ATM approach in comparison to traditional system integration approaches in the ATM domain.

Abstract. The capability to provide a platform for flexible business services in the Air Traffic Management (ATM) domain is both a major success factor for the ATM industry and a challenge to integrate a large number of complex and heterogeneous information systems. Most of the system knowledge needed for integration is not available explicitly in machine-understandable form, resulting in timeconsuming and error-prone human integration tasks. In this chapter we introduce and evaluate a knowledge-based approach, "Semantically-Enabled Externalization of Knowledge" for the ATM domain (SEEK-ATM), which explicitly models a) expert knowledge on specific heterogeneous systems and integration requirements; and b) allows mapping of the specific knowledge to the general ATM problem domain knowledge for semantic integration. The domain-specific modeling enables a) to verify the integration knowledge base as requirements specification for later design of technical systems integration and b) to provide an application program interface (API) to the problem space knowledge to facilitate tool support for efficient and effective systems integration. Based on an industry case study, we evaluate effects of the proposed SEEK-ATM approach in comparison to traditional system integration approaches in the ATM domain. Major advantages of the novel approach are the efficient derivation of technical configurations and automated quality assurance of the expert knowledge models.

Database technologies are evaluated in respect to their performance in model extension, data integration, data access, querying and distributed data management. The structure of the data sources is partially unknown. Additional value is gained combination of data sources. Data models for a relational, a document and a graph oriented database are compared showing strengths and weaknesses of each data model.