An Analysis of Content Gaps Versus User Needs in the Wikidata Knowledge Graph

David Abián, Albert Meroño-Peñuela, Elena Simperl

Abstract: Content gaps in knowledge graphs impact downstream applications. Semantic Web researchers have studied them mainly in relation to data quality or ontology evaluation, for instance by proposing frameworks to capture various quality dimensions or methods to assess these dimensions, such as completeness, accuracy, or consistency. Less work has been done in framing these gaps in the context of user needs. This limits our ability to design processes and tools to help knowledge engineers tackle such gaps effectively. We propose a framework that: (i) captures core types of content gaps, informed by a literature review on peer-production systems; and, in the areas with such gaps, (ii) quantitatively compares the imbalances in the work on the knowledge graph with the imbalances in users’ information needs to clarify the origin of the gaps. We operationalize the framework with gender, recency, geographic, and socio-economic gaps, and apply it to Wikidata by comparing edit metrics with Wikipedia pageviews between 2018 and 2021. We did not find gender or recency gaps endogenous to Wikidata’s production. Only exceptionally, Wikidata editors work on under-represented entities (e.g. people from countries with lower Human Development Index) less than they should according to the volume of requests. We hope this study will provide a foundation for knowledge engineers to explore the causes of content gaps and address them if and when needed.

External IDs:dblp:conf/semweb/AbianMS22