Quantitative assessment of metadata collections of research data repositories

Abstract

Structured metadata are of particular importance in the context of facilitating research data (re-)use. Although research data repositories create and manage metadata records, existing research offers limited insights into the relationship between repositories and metadata for research data. Therefore, in conducting a quantitative assessment informed by metadata quality requirements, this thesis aims at making distinctive features of metadata for research data visible, specifying the potential influence of repository characteristics on metadata, and exploring changes to metadata records. The analysis showed variations in metadata completeness across repositories. Within repositories, metadata descriptions are relatively homogenous. These findings suggest that repositories have developed distinctive and consistent practices for describing data. On average, descriptions comprise 487.3 characters, and 5.52 years passed between the year a dataset was published and the metadata record was registered. Differences in the completeness of metadata records, description length and timeliness were significant across repository types and certification status, whereas differences in collection homogeneity were not significant. Overall, most metadata records in the sample were changed, which conforms with the conceptualization of metadata for research data as dynamic and changeable objects. Differences in the number of changes are significant across repository types.