Methodology

In order to address the problems outlined above, we define a number of potential quality issues that, after human judgement, help to identify errors and misconceptions in a given vocabulary. For each issue, we give a real-world example and outline the actual implementation. We grouped these criteria into:

• '''Labeling and Documentation Issues'''
• '''Structural Issues''': quality issues that can be derived from the graph-based nature of SKOS vocabularies, such as missing or misplaced relations between concepts
• '''Linked Data Specific Issues''': quality issues that can be derived from the Linked-Data approach of organizing data on the Web

We also set up a [https://github.com/cmader/qSKOS-data github repository] containing the set of SKOS vocabularies we have already tested against the identified issues. The dataset is expected to grow as qSKOS development proceeds.