GOLDCASE: A Generic Ontology Layer for Data Catalog Semantics

Data catalogs automatically collect metadata from distributed data sources and provide a unified and easily accessible view on the data. Many existing data catalog tools focus on the automatic collection of technical metadata (e.g., from a data dictionary) into a central repository. The functionality of annotating data with semantics (i.e., its meaning) in these tools is often not expressive enough to model complex real-world scenarios. In this paper, we propose a generic ontology layer (GOLDCASE), which maps the semantics of data in form of a high-expressive data model to the technical metadata provided by a data catalog. Hence, we achieve the following advantages: 1) users have access to an understandable description of the data objects, their relationships, and their semantics in the domain-specific data model. 2) GOLDCASE maps this knowledge directly to the metadata provided by data catalog tools and thus enables their reuse. 3) The ontology layer is machine-readable, which greatly improves automatic evaluation and data exchange. This is accompanied by improved FAIRness of the overall system. We implemented the approach at PIERER Innovation GmbH on top of an Informatica Enterprise Data Catalog to show and evaluate its applicability.