Dynamic Spreadsheet Editor for Knowledge Graphs based on Shape Constraints

Abstract

Knowledge Graphs (KGs) have become increasingly more important in recent years. As structured representations of knowledge, they enable the semantic connection and integration of large amounts of data and provide a powerful basis for data-driven applications. The growing relevance of KGs is not only reflected in industrial applications, but also in academic research, where they are increasingly understood as a key element of modern information systems. One problem in the practical use of KGs is access to the data for non-experts in the field. However, domain experts are particularly important in the further development, maintenance and correction of KGs. This thesis aims to develop a tool that closes the gap between knowledge graph users, who are non-experts in this field, and ontology engineers. In this thesis, ShaclSheet, a JavaScript-based web application, was developed that can generate fully editable tables for a knowledge graph. The table structure is automatically generated by a set of configurations created in advance by Ontology Engineers. The configuration of the tables is based on the constraint language SHACL, which was developed as a validation tool for RDF graphs and offers many possibilities for generating editors. In the course of this development, a comprehensive requirement analysis was carried out. This was used as the basis for the design and subsequent development of the editor. The central components and functionalities of the editor are presented. This was followed by a user evaluation of the editor within DLR. This focussed on the simplification and time savings of working with ShaclSheet compared to working with SPARQL queries. For this purpose, tasks were to be completed in both cases and then the difficulty of the respective tasks and the experience of working with ShaclSheet were to be specified in a survey. The evaluation was carried out both quantitatively by measuring the time taken to complete the tasks and analysing the rating on a Likert scale, and qualitatively by analysing the comments made by the test subjects. Finally, an evaluation of the development of ShaclSheet and an outlook on further questions were given.