Schema and Ontology Matching
Schema / ontology matching is the task of finding semantic correspondences between elements of two schemas or two ontologies. It is needed in many metadata-intensive applications, such as integration of web data sources, catalog integration/merging, data warehouse loading, XML message mapping and peer-to-peer data management. Currently, such matching tasks are largely performed manually by domain experts, and therefore they are time-consuming, tedious and error-prone. Approaches for automating the schema and ontology matching tasks as much as possible are needed to simplify and speed up the development, maintenance and use of metadata-intensive applications.
Our contributions to semi-automatic schema and ontology matching are as follows:
- Develoment of a solution taxonomy of the major approaches to semi-automatic schema matching (VLDB Journal 2001, VLDB 2001)
- Development of a new graph-based structural matcher called Similarity Flooding (Best Student Paper, ICDE 2002). A prototypical implementation is available including source code.
- Development of the COMA and COMA++ platforms to combine different match approaches in a flexible way and to reuse previously determined match results (VLDB 2002,
SIGMOD2005,
IS2007).
- Comprehensive evaluation of different match techniques based on precision, recall and a new combined metric, called Overall (ICDE 2002, VLDB 2002, LNCS2593, IS2007))
- Investigation of instance-based methods for schema and ontology matching and their succesful use for solving large-scale match tasks (e.g., matching of life science ontologies, product catalogs, and web directories).
Publications
|
