Evaluation of metadata in research data repositories: The case of the DC.Subject element

Abstract

Research Data repositories are growing in terms of volume rapidly and exponentially. Their main goal is to provide scientists the essential mechanism to store, share, and re-use datasets generated at various stages of the research process. Despite the fact that metadata play an important role for research data management in the context of these repositories, several factors - such as the big volume of data and its complex lifecycles, as well as operational constraints related to financial resources and human factors - may impede the effectiveness of several metadata elements. The aim of the research reported in this paper was to perform a descriptive analysis of the DC.Subject metadata element and to identify its data quality problems in the context of the Dryad research data repository. In order to address this aim a total of 4.557 packages and 13.638 data files were analysed following a data-preprocessing method. The findings showed emerging trends about the subject coverage of the repository (e.g. the most popular subjects and the authors that contributed the most for these subjects). Also, quality problems related to the lack of controlled vocabulary and standardisation were very common. This study has implications for the evaluation of metadata and the improvement of the quality of the research data annotation process.