Comparing semantic matching results of schema matchers and metadata registry enabled systems

The goal of schema matching is to find semantic correspondences between two data sources to establish interoperability. This work proposes that the semantic descriptions of data elements stored in metadata registries can be used to establish semantic interoperability between database schemas more effectively than direct schema-to-schema matching because metadata registries contain more detailed semantic information than it is possible to glean from database schemas. The schemas of two protein databases are matched using both a metadata registry assisted approach and an automatic schema matcher. The resulting suggested matches from both methods are compared and the possible reasons for superior metadata registry assisted matching results are discussed.