Choose your own adventure to a reproducible scientific article: learnings from ReproHack
Bibliographic record
Abstract
Toronto Data Workshop on Reproducibility - 26 February 2021 "I shared the code and data of my last scientific article, does it mean that it is reproducible?" We could think that having access to the research data and code would be enough to reproduce published results, but often this is much more involved. Is reproducibility dependent on the reviewer's knowledge? What things do we not usually think about can affect reproducibility? Can the choice of how to capture the computational environment influence the experience of the reviewer? In this talk, we are going to think together some of the necessary steps that make someone else able to reproduce a scientific article or project. I will share some thoughts from my experience in ReproHack and show you how reviewing is a great practice to learn about reproducibility. What is ReproHack? Reprohack is a hackathon-style event focused on the reproducibility of research results. These hackathons provide a low-pressure sandbox environment for practicing reproducible research: Authors can practice producing reproducible research and receive friendly feedback and appreciation of their efforts Participants can practice reviewing, learn about reproducibility best practices as well as common pitfalls from working with real-life materials rather than just dummy. They also get inspired and grow confidence in working more openly themselves. Research Community benefits from: Evaluating what best practice is in practice More practice in both developing and reviewing materials Illustrations The Turing Way Community, & Scriberia. (2020, March 3). Illustrations from the Turing Way book dashes. Zenodo. http://doi.org/10.5281/zenodo.3695300 Links to the slides: bit.ly/reprohack_toronto https://flor14.github.io/toronto_data_workshop/what_i_learned_from_reprohack.html
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 | 0.171 | 0.436 |
| Meta-epidemiology (narrow) | 0.001 | 0.002 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.003 | 0.003 |
| Science and technology studies | 0.011 | 0.020 |
| Scholarly communication | 0.021 | 0.021 |
| Open science | 0.005 | 0.020 |
| Research integrity | 0.007 | 0.019 |
| Insufficient payload (model declined to judge) | 0.031 | 0.038 |
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; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
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".