Involving Data Creators in an Ontology-Based Design Process for Metadata Models

Introduction

As the research environment is increasingly driven by data, research data management is gradually becoming a very important requirement for research projects. In the absence of proper management, expensive and irreplaceable research data may never realize their reuse potential; at the same time, their availability usually declines steadily as the publications age (Vines et al., 2014). While this is a problem for large-scale projects, it is even more prevalent in the context of research groups, or single researchers, in the long-tail of science (Heidorn, 2008), that often operate with very limited resources to ensure sustainability of their data.

To deal with this pressing issue, an increasing number of research funders are demanding research grant applicants to include data management plans in their project proposals - especially if public funds are required. These data management plans state, among other things, where and how the data will be deposited, preserved and kept accessible after the formal conclusion of the project. Major research funding providers are demanding such data management plans with recent calls for projects – examples include the European Commission under Horizon2020 (European Commission, 2013), and the National Science Foundation, in the US (National Science Foundation, 2011). Some publishers have also started to request data as supplementary materials to the submitted articles, under the assumption that their readers should be able to validate or replicate the presented results. Nature, for instance, requests authors to disclose research materials as a condition for the publishing of research papers. Another example is an Open Access publisher, PLOS ONE, that demands a full, unrestricted access to the original data for each of the submitted manuscripts. Following these trends data management is already an important concern for the scientific community.

The investment in research data management is important for many reasons: not only does it improve the chances of reproducibility and verifiability of the research results but can also prevent fraud. Another advantage of promoting data reuse relies on decreasing data duplication and the inherent research efforts to produce them. This allows researchers to directly focus their work in the project's specific goals, leaving more time to pursue an extensive validation or other research activities.

Research data management workflows involve both practical issues faced by the process stakeholders as well as technical ones. Sound technological solutions to support institutional repositories have been presented to reduce the technical issues, and we have seen great progress in that regard; solving the practical issues is, however, a challenge that is far from being settled, as it depends on fostering the interest of researchers to be active stakeholders in the data management workflow, more precisely in the description of their data. Data description assumes a critical nature in this workflow as it enables researchers with interest in a dataset to find and reuse it. Thus, the dissemination and preservation of research data strictly relies on metadata (Treloar and Wilkinson, 2008). Practical and technical issues are therefore related, and the need for high-quality metadata drives the technical developments often seen in research data management infrastructures.

However, data description is very demanding and time consuming, let alone the research process itself, so researchers are progressively investing in metadata creation, dealing with it during their daily activities.

This involves producing data from diverse sources and extracting their production context that is often kept in laboratory notebooks. By nature, such records follow an unstructured approach, strongly dependent on the researchers’ perspective. When this is the case, researchers generate metadata that may lose their value upon the project's closure, as their interpretation can be problematic for external parties. In fact, an international survey demonstrate that researchers agree on the satisfaction of data saving in the short-term and the capacity to analyse the data they are creating (more than 70 per cent); however, the same does not apply for storing the data beyond the lifetime of a project, as only 44 per cent of them were satisfied with the access to long-term research data (Tenopir et al., 2011).