Recover a published curve
Use the Paper-to-Benchmark Builder to calibrate a source figure, trace experimental data and preserve the extraction record.
LingoScholar · Research methods
Reproducibility LabMake every assumption visible.
Bring experimental evidence, explicit models and repeatable comparisons into one research workflow.
Free for approved academic accounts. Use your LingoScholar email and password.
01 / EvidenceGeometry · loading · sourcesMeasured, digitized or derived
02 / ModelEquations · parameters · unitsAssumptions and calibration stated
03 / EvaluationCampaign split · errors · scopeA repeatable record of the result
Versioned data. Traceable assumptions.
Available evidence
7 source-checked specimen records · 1 campaign
The current public collection is available in the specimen atlas. Models begin with labelled, editable demonstration inputs. Source-validated opening and VOL reference models still require curation.
Make the model explicit
Edit effective wall or beam properties, inspect the equations and rerun the same idealized model in Python.
Separate fitting from evaluation
Compare model predictions and hold out whole experimental campaigns. Inspect error and coverage for each model.
Plan a discriminating experiment
Rank proposed tests by disagreement between your model predictions, with feasibility and evidence gaps visible.
Keep a reproducible snapshot
Download source-linked specimen data and preserve model sessions, assumptions and calculation reports.
Contribute with review
Submit experimental or model records for owner review. Approval and public release are separate decisions.
From figure to dataset
Paper-to-Benchmark Builder
Trace an experimental figure with calibrated axes, record its measurement definitions and compare your prediction. Preserve the source image and extraction choices in a reusable package.
Cross-lab comparisons
Timber Benchmark Exchange
Download a fixed experimental case, submit your own model predictions and compare reviewed methods using the same displacement domain and measurement definitions.
Researcher contributions
Reviewed source records
Published metadata credits its contributor and links to the original source. Editorial review does not establish scientific validity.
No contribution metadata has been published yet. Approved academic members can submit records through the research workspace.
Submit a documented contributionA clear record
Evidence and predictions keep their identities.
Source-linked data packages preserve measurement definitions and reuse licences. Model sessions retain declared parameters. Comparison reports distinguish calibration campaigns from held-out campaigns.
- Unknown inputs stay unknown.
- Demonstration numbers are labelled synthetic.
- Opening effects and failure sequences are never inferred from missing data.
- Your model files are processed locally; contributions are sent only when you submit them.