Towards an Architecture to Support Data Access in Research Data Spaces

This is the authors' abstract. We don't add key points for this paper.

Using data from different data sources is a common procedure in data-driven research. As required data is often not available from centrally managed sources, the concept of data spaces has more and more frequently been utilized to integrate decentral data sources and supporting the access to these. However, decentrally organized data management leads to differences in data models, formats and technologies and the problem of matching data from different sources most often still requires significant manual work, either from the provider or the consumer of the data. Especially in research data spaces, properties of data sources are often dynamically modified or extended, and complex queries are formulated to access the data. In this paper, we present an architecture to approach this problem by automatically exploring the different data sources and searching for similar attributes. Matching attributes are then assembled to so-called vocabularies, that represent common concepts in the data space independently from their actual representations. This dynamic knowledge base is then used to efficiently query and access data in the data space. Finally, an application of the concept is presented to demonstrate the applicability of our architectural framework.