In Proceedings of the 4th International Workshop on Visualization and Interaction for Ontologies and Linked Data (2018)
Data quality is an important factor for the success of the envisaged Semantic Web. As machines are inherently intolerant at the interpretation of unexpected input, low quality data produces low quality results. Recently, constraint languages such as SHACL were proposed to assess the quality of data graphs, decoupled from the use case and the implementation. However, these constraint languages were designed with machine-processability in mind. Defining data shapes requires knowledge of the language's syntax - usually RDF - and specification, which is not straightforward for domain experts, as they are not Semantic Web specialists. The notion of constraint languages is very recent: the W3C Recommendation for SHACL was finalized in 2017. Thus, user interfaces that enable domain experts to intuitively define such data shapes are not thoroughly investigated yet. In this paper, we present a non-exhaustive list of desired features to be supported by a user interface for editing data shapes. These features are applied to unSHACLed: a prototype interface with SHACL as its underlying constraint language. For specifying the features, we aligned existing work of ontology editing and linked data generation rule editing with data shape editing, and applied them using a drag-and-drop interface that combines data graph and data shape editing. This work can thus serve as a starting point for data shape editing interfaces.