MERMaid: Universal multimodal mining of chemical reactions from PDFs using vision-language models
Bibliographic record
Abstract
Data digitisation of scientific literature is essential to expedite the creation of machine-learnable knowledge bases for data-driven research and integration with knowledge-intensive systems like self-driving laboratories. However, automating the extraction, interpretation, and the structuring of data from information-rich graphical elements within the prevalent PDF format remains a significant challenge. We present MERMaid (Multimodal aid for Reaction Mining), an end-to-end knowledge ingestion pipeline to automatically convert disparate information conveyed through figures and tables across various PDFs into a coherent and machine-actionable knowledge graph. By leveraging the emergent visual cognition and reasoning capabilities of vision-language models, MERMaid demonstrates chemical context awareness, self-directed context completion, and robust coreference resolution to achieve 87% end-to-end overall accuracy. Notably, MERMaid is topic-agnostic and adaptable to various chemical domains. Its modular design and extensibility facilitate future application to diverse scientific data beyond reaction mining, promising to unlock the full potential of scientific literature for knowledge-intensive applications.
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 | 0.001 | 0.006 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.001 | 0.002 |
| Bibliometrics | 0.003 | 0.001 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.003 | 0.003 |
| Open science | 0.002 | 0.003 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.008 | 0.005 |
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".