How To Write A (Good) Data Description: Developing Best Practice
Bibliographic record
Abstract
When researchers share their research as a published journal article, it includes an abstract. We understand these well: we don't always follow best practices, but there is substantial literature on what should be in an abstract, guides to writing effective abstracts, and journal-level standards for highly structured abstracts. The same is not true for descriptions of published data sets, and the advice for journal articles is not universally applicable. Researchers are increasingly incentivized to make research data available in a public data repository, documented using controlled vocabularies and well-defined metadata fields. Most data documentation standards include a title and at least one field for summarizing the data set which are open-ended, plain text, unconstrained fields. These "unmanaged” fields are particularly important for most search engines, as they are the metadata fields most consistent with natural language, for which search has been highly optimized. They play an important role in some search and filtering tasks, particularly among emerging scholars and novice users. While expert data managers and others have developed the ability to write thorough and useful dataset descriptions, we've observed that data repositories and catalogues, and most data documentation standards, have inconsistent and vague instructions on the content of the dataset description field. Broadly, our objective is to establish evidence-based guidance for effective documentation of datasets using unstructured text fields. We have reviewed existing literature and best practices to establish core guiding principles to support the authors of dataset descriptions. These principles have been refined through consultations with data librarians, data repository managers, and other experts. This poster describes the refined proposed guidelines, explains the reasoning behind each, and solicits input and feedback on what is required for a set of guiding principles to help users write better, more useable data set descriptions.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.553 | 0.732 |
| Meta-epidemiology (narrow) | 0.003 | 0.007 |
| Meta-epidemiology (broad) | 0.004 | 0.004 |
| Bibliometrics | 0.034 | 0.026 |
| Science and technology studies | 0.013 | 0.032 |
| Scholarly communication | 0.045 | 0.064 |
| Open science | 0.018 | 0.028 |
| Research integrity | 0.016 | 0.027 |
| Insufficient payload (model declined to judge) | 0.007 | 0.019 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; the direct Gemma label and the distilled Codex classifier agree on what is shown here.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".