Springe direkt zu Inhalt

Jakob Warkotsch:

Screenshot of Visual Query Builder

Screenshot of Visual Query Builder

Developing a Graphical User Interface for Generating Wikipedia Lists with Wikidata

Master of Science (M.Sc.)
Jan 05, 2018


Wikipedia is one of the most popular websites on the World Wide Web, pro- viding free knowledge created by a community of volunteer editors. It is sup- ported behind the scenes by Wikidata, the free and open knowledge base, another community-based Wikimedia project, which is a platform for the cre- ation and management of structured data. Wikidata can be used to enrich Wikipedia articles with information and is especially useful outside of con- tinuous text, for the display of tabular data or lists. At present, there are only programmatic solutions for the creation of lists using Wikidata’s data. This thesis with the title ”Developing a Graphical User Interface for Generat- ing Wikipedia Lists with Wikidata” is about the development of a tool which aims to enable experienced Wikidata users to produce SPARQL queries for automated list generation without having to know SPARQL.

An extensive assessment of existing graphical SPARQL query builders was conducted, building upon an existing categorization to propose a hierarchical systematization of such tools based on their primary purpose and their user interface paradigm in order to learn from existing approaches and the possible trade-offs between simplicity and functionality. To find out which features of SPARQL and which graph patterns for queries needed to be supported by the tool, an analysis was performed on queries produced by the English and German Wikipedia community. Combining the insights from the analysis phase, a conceptual model was created with a novel mapping of Wikidata concepts to SPARQL constructs. A user-centered design approach was taken to iteratively test and improve the design with potential users with a low- fidelity and a high-fidelity prototype.

A functional version of the high-fidelity prototype was tested in a user study at the WikidataCon. Qualitative feedback was gathered using the thinking aloud method during the test sessions, and a standardized questionnaire was used to generate quantitative data, which also allowed the Query Builder’s usability to be evaluated against the questionnaire’s benchmark. Both the questionnaire results and the personal feedback at the conference suggested that the Wikidata Query Builder is going to be a useful and usable tool.