Model based research data management

Abstract

Open Science aims to improve reproducibility and reuse by making research more accessible and transparent. A key aspect of Open Science is enhancing publications with Open Data (i.e., sharing research data, the foundation of scientific knowledge). So-called experiment model based research data management systems (ERDMSs) base the documentation of the stored research data on domain-specific experiment models that describe the data in greater detail. Embedding an explicit experiment model into research data management supports researchers in describing and understanding the data stored in a system. However, ERDMSs require more documentation effort from the researchers to enhance their research data with additional information. Standard data forms, like spreadsheets, are typically not sufficient to describe sophisticated, multi-step processes, such as the design of a research experiment. As such, well-designed documentation forms representing individual research workflows are essential to increase the usability of ERDMSs. However, the implementation and use of ERDMSs is a major challenge. The underlying model must be initialized and maintained, which is time-consuming, expensive, and leads to high operating costs.

Through eight quantitative and qualitative studies, this thesis demonstrates how research data management can be integrated into everyday research using ERDMSs and how they support scientists in the concurrent documentation of research data. For this purpose, this thesis presents new methods for structured data entry, adaptability of data documentation, data exploration, experiment model initialization, and maintenance. These methods could improve the data documentation significantly in terms of both in usability and in the perceived effort. Simultaneously, the technical effort needed to maintain the model could be significantly reduced by more closely involving the domain experts in material sciences in the model maintenance process.