Which attributes may affect the retrieval rate of your data? Evidence from an investigation into a large international data portal

Open science knowledge has significant value, yet this value can only be realised if the knowledge is accessible, findable and retrievable. Although it is well-documented that detailed metadata are important, the more critical point is what factors are more important and worth of more investments and managerial efforts. We collect data from a major Australian science repository and show that: compared to traditional factors, including metadata richness, contributor network and reputation, age, and fields of research, the topicality of datasets’ keywords has a more significant power deciding whether the datasets become top-downloaded datasets. Among traditional factors, metadata completeness has the strongest positive effect. We thus provide theoretical and practical insights about how to increase retrievals, unlocking greater value for users.