A comparison is only as clear
as the record behind it.

Define the review before interpreting it

Save a focused question, population, inclusion and exclusion criteria, search strategy and date, outcome and its definition, and the condition you intend to compare. The recorded search describes the work you performed; the Studio does not automatically search for new publications or watch sources continuously.

The current outcome types cover force, stiffness, stress and displacement. A title such as “stiffness” is insufficient on its own: record the measured response, displacement location, force range and calculation convention where relevant. A different quantity requires a separate review or an explicit, defensible transformation.

Each record keeps its inclusion, exclusion or uncertain decision and the reason. An unresolved record remains visible without being silently admitted to a calculation. Conclusions and unresolved questions are written by the researcher.

Keep the value and its source together

A record retains the paper identifier or citation, DOI or source URL where available, exact page, table or figure locator, source context, specimen, campaign, numerical value, units and outcome definition. Record extraction and calculation limitations, including information missing from the source.

“Source checked” means the contributor states that they checked the recorded value and locator against the source. It does not imply independent scientific validation, a complete audit of the paper, or permission to redistribute its contents. You remain responsible for citation, reuse permissions and the interpretation of the original study.

Recognize study families and shared specimens

Give records from the same experimental campaign a consistent campaign key, and keep specimen identifiers consistent across papers. The family view groups declared campaign records, specimens and source papers so repeat reporting stays visible. Multiple articles on the same specimen or campaign do not become additional independent experiments.

Campaign keys and specimen identities are contributor declarations. Different keys do not verify independence: campaigns may share a laboratory, material batch, team, test setup or model calibration data. Missing identity information remains unknown. The family view does not automatically discover duplicate experiments or prove independent replication.

Compare conditions conservatively

The protocol names one factor of interest. The comparison reviews the other recorded conditions rather than treating every difference as evidence of that factor's effect. A match means agreement in the recorded fields; it does not establish physical equivalence.

• Material, element, lamination orientation, lay-up, specimen dimensions and opening, lintel and parapet geometry.
• Connections, boundary conditions, loading protocol and direction, and applied gravity loading.
• Displacement measurement definition and stress-area basis, including whether a stress uses gross or net area.
• Species, grade and moisture condition where relevant to interpretation.

Blank or unknown information is a blocker, not an agreement. Where a condition genuinely does not apply, an explicit “not applicable” statement with its context is supported; it is not a substitute for missing information. A mismatch also remains a blocker unless the researcher records a permitted field-specific override with a substantive reason. Any such override remains visible and labels the contrast conditional. An override does not transform the source value, correct an incompatible outcome definition or establish causality.

The stress-area basis is a critical condition for stress outcomes. The displacement measurement definition is critical for stiffness and displacement outcomes. Missing or incompatible critical definitions block the calculation even when that field is selected as the factor of interest or supplied with an override. Invalid physical numerical conditions also block the calculation and cannot be overridden.

The calculation also checks record eligibility, source completeness, checking declarations, origin, quantity definitions, supported units and identifiable campaign/specimen information. Hard eligibility issues cannot be removed with a condition override. Review every issue and every source before interpreting a contrast.

Descriptive contrasts and unit handling

A comparison uses an explicitly selected baseline A and comparison B. Compatible supported units are normalized as listed below. Unit labels are case-sensitive. Values with unsupported units or incompatible quantity definitions are withheld from the calculation. The original source values and units stay in the dossier.

A zero baseline leaves the ratio and percentage change undefined; no substitute denominator is introduced. Eligible conditional contrasts retain their issues and reasons beside the result. An unavailable result remains unavailable rather than being replaced with zero.

These are descriptive contrasts between selected records. They do not estimate a causal effect, adjust for confounding, provide a pooled material property, or establish statistical significance. A single campaign does not support a cross-laboratory consensus. The Studio does not calculate a universal evidence score, confidence interval or pooled meta-analysis.

Import deliberately and retain the boundary

Selected curated Atlas records can provide source-linked starting data. Selected evidence from a Research Room can provide a starting review record when your account has access. Review imported values, units, identities and conditions; an import does not fill unknown information or establish eligibility.

Imported evidence is a copy with its recorded reference and revision. Changes to the source record are not automatically reconciled into your review. The Paper-to-Benchmark Builder documents local curve extraction, and the Benchmark Exchange retains fixed-case model comparisons; their results do not become experimental review evidence automatically.

Freeze a version and export the dossier

Saving updates the working project. Freezing deliberately creates a retained snapshot of that saved revision, including its protocol, sources, values, conditions, decisions, comparisons, computed analysis, conclusion and unresolved questions. Later edits do not replace an existing frozen version.

Export a frozen dossier as JSON, data tables as CSV, and a printable HTML report. Keep the source locator and limitations with any exported result. The snapshot records its freeze time, project revision, method version and SHA-256 fingerprint. Verification checks that the dossier contents match the recorded fingerprint.

A matching fingerprint establishes copy integrity relative to that fingerprint. It does not prove the source data are true, the interpretation is correct, the work is independently validated, or an external party certified the record. This is a deliberately frozen review version, not an automatically maintained living search.

Current data and privacy

The current curated Atlas contains seven solid-wall specimens from one campaign. They can support within-campaign descriptive reviews. Opening and VOL questions require additional curated data; the Studio must retain “insufficient comparable evidence” when the selected records do not support a comparison.

Projects are private to your LingoScholar account. Service administrators can access stored data. Exporting or deliberately sharing a dossier creates a copy that cannot be recalled. Private owner PDFs are not imported automatically, and creating a review does not publish it or invite other researchers.

Use your existing LingoScholar email and password. The workspace is available to approved academic accounts and the website owner.