A Framework to Assess Knowledge Graphs Accountability

TL;DR

This paper introduces KGAcc, a framework for assessing the accountability of Knowledge Graphs, focusing on accountability requirements and measures, with evaluations on Linked Open Data and comparisons to existing frameworks.

Contribution

The paper presents KGAcc, a novel framework specifically designed to evaluate the accountability of RDF graphs, filling a gap in existing data quality assessments.

Findings

KGs from the LOD cloud vary significantly in accountability levels.

The KGAcc framework effectively differentiates between more and less accountable KGs.

Comparison shows KGAcc offers unique insights not covered by existing data quality frameworks.

Abstract

Knowledge Graphs (KGs), and Linked Open Data in particular, enable the generation and exchange of more and more information on the Web. In order to use and reuse these data properly, the presence of accountability information is essential. Accountability requires specific and accurate information about people's responsibilities and actions. In this article, we define KGAcc, a framework dedicated to the assessment of RDF graphs accountability. It consists of accountability requirements and a measure of accountability for KGs. Then, we evaluate KGs from the LOD cloud and describe the results obtained. Finally, we compare our approach with data quality and FAIR assessment frameworks to highlight the differences.