A Web-based UI to Enable Semantic Modeling for Everyone

André Pomp, Alexander Paulus, Daniel Klischies, Christian Schwier, Tobias Meisen

Abstract: Since companies generate and store large amounts of data daily in centralized systems such as data lakes, understanding data sets from different sources is becoming an increasingly complex task in dealing with data heterogeneity across domains. One solution for describing semantics of data sources is the use of semantic models based on an available vocabulary. However, creating detailed semantic models can be a challenging task for users who are not familiar with semantic modeling and today’s available tools. To overcome this challenge, we developed an intuitive and user-friendly interface, allowing data owners to define detailed semantic models for their data sources. The design of the user interface is based on an intensive requirement analysis gathered among several peers. It provides an intuitive mapping of semantic concepts to data attributes and the definition of relations between those concepts using drag and drop interaction. The user is given full modeling freedom as the insertion of semantic concepts and relations that are missing in the underlying vocabulary can be done on-demand and does not delay or impair the modeling process. Additionally, the refinement of the original detected data schema is supported with several operations. We built the interface into the semantic data platform ESKAPE, which already uses a flexible knowledge graph as underlying vocabulary and provides a detailed analysis of the data schema.