Process-based schema matching: from manual design to adaptive process construction

Mappings between complex metadata structures are needed in a number of domains such as data integration, ontology alignment or web service composition. A mapping describes how elements of one metadata structure correspond to elements of another metadata structure. Defining such mappings is a complex and time-consuming process. It is often done manually, with the help of point and click interfaces. In the last 10 to 15 years a strong effort was made in research to partly automate the mapping process. Many schemaand ontology matching systems were developed to semi-automatically compute mapping suggestions for a user. Current systems are often not robust enough to be able to cope with different mapping problems and often face performance issues when mapping large structures. This thesis investigates how to support the user in the task of configuring schema matching systems and how to improve run-time performance.