Search

Researchers are increasingly required to make research data publicly available in data repositories. Although several organisations propose criteria to recommend and evaluate the quality of data repositories, there is no consensus of what constitutes a good data repository. In this paper, we investigate, first, which data repositories are recommended by various stakeholders (publishers, funders, and community organizations) and second, which repositories are certified by a number of organisations. We then compare these two lists of repositories, and the criteria for recommendation and certification. We find that criteria used by organisations recommending and certifying repositories are similar, although the certification criteria are generally more detailed. We distill the lists of criteria into seven main categories: “Mission,” “Community/Recognition,” “Legal and Contractual Compliance,” “Access/Accessibility,” “Technical Structure/Interface,” “Retrievability” and “Preservation.” Although the criteria are similar, the lists of repositories that are recommended by the various agencies are very different. Out of all of the recommended repositories, less than 6% obtained certification. As certification is becoming more important, steps should be taken to decrease this gap between recommended and certified repositories, and ensure that certification standards become applicable, and applied, to the repositories which researchers are currently using.

Corpus studies suggest that verb tense is a differentiating feature between, on the one hand, text pertaining to experimental results (involving methods and results) and on the other hand, text pertaining to more abstract concepts (i.e. regarding background knowledge in a field, hypotheses, problems or claims). In this paper, we describe a user experiment that investigates whether for biological readers, this tense correlation has a psychological correlate. To study this, we defined seven distinct discourse segments types and modified them either by changing the verb tense/mood (for all segment types), negation (for Problems), or presence of an epistemic matrix clause (‘These results suggest…’) for Implications. Regardless of the original segment type, we found that for Facts, Results and Hypothesis segments, present tense yielded more Fact classifications, past tense more Result interpretations, and modal auxiliaries more Hypothesis interpretations. Methods statements were less sensitive to verb form. Problem segments required negations to be recognized, while Implications required introductory matrix clauses.