HELP

Glossary

The words EasyQLab uses on screen and in these docs, in alphabetical order. Each one links to the page that explains it in full.

A – C

Analysis

One pass of the checks over an original and its translations. It is saved inside a project, numbered, and doesn’t change once it is final. Also called a run. See The results screen.

Annotated PDF

The translated PDF with every finding that counts added as a standard PDF comment, plus a summary at the end. The content of the page is not touched. See Reports and exports.

Approved

The green verdict: no critical or high findings, and fewer medium ones than the threshold. See The traffic light.

Category

The family a check belongs to, such as Layout, Punctuation or Translation. Findings can be filtered and sorted by it. See What it checks.

Check

One of the 47 tests EasyQLab runs on every page, each with a code (C4, C39, X1…) and a default severity. See What it checks.

Client list

A CSV you upload in Settings → Client lists with a client’s font equivalences, glossary, forbidden terms, non-translatables and brands. See Settings.

Confidence

How sure EasyQLab is that a finding is a real problem. Findings below the cut-off are hidden and don’t count. See The traffic light.

Consensus check

A check that compares the languages with each other and points at the one that differs from all the rest (X1, X2, X3). It needs three languages or more. See What it checks.

Cut-off

The confidence from which a finding counts: 60 % by default, 50 % for linguistic findings. Below 40 % nothing is kept. See Confidence.

D – L

Doubtful

A finding between 40 % confidence and the cut-off. It is hidden and doesn’t count until someone confirms it with Real error. See Hidden findings.

DTP

Desktop publishing: laying out the translated text in the final design. Many findings are for the DTP operator, not the translator.

Finding

One problem EasyQLab found, marked on the page, with its severity, check, page and explanation. See The finding card.

Font equivalence

A font you authorise to replace another in a given language. Authorised replacements don’t raise a finding. Set them in Settings → Typefaces. See Settings.

Human review comparison

Uploading the PDF your reviewer returned, with their comments, to see what both found, what EasyQLab missed and what only EasyQLab flagged. See Second round and human review.

Ignored

A finding you took out of the count with ignore. It leaves the traffic light and the annotated PDF, and stays in the CSV marked “ignored”. See Ignoring a finding.

Incomplete

A project in which at least one language could not be analysed. It can’t be approved until that language is run again. See Incomplete projects.

Language card

The box for each language at the top of the results, with its number of findings, verdict and severity bars. See The language cards.

Layout finding

Any finding that is not linguistic: format, layout, numbers, images. Counted in the Layout column of the Summary. See The Summary tab.

Linguistic finding

A finding about the words: the translation review, spelling and forbidden terms. It has its own confidence cut-off and its own column in the Summary. See Confidence.

M – Q

Major

A finding of severity 4 (high) or 5 (critical). The card says “N major” or “no major”. See Severities.

Mirrored layout

A right-to-left design that flips the page, as is usual in Arabic or Hebrew. Images that moved because of mirroring are expected. See Right-to-left languages.

Needs review

The amber verdict: at least one critical or high finding, or too many medium ones, but not enough to reject. See The traffic light.

Not an error

The button for a false alarm. The finding is dismissed, stops counting, and EasyQLab lowers its confidence in that kind of finding for your next projects. See Deciding.

Original (Source)

The document in the source language, usually English, that every translation is compared with. On the upload screen it is the Source box.

Overlay

The viewer mode that lays the translation over the original, with an opacity slider and a Difference option. See The viewer.

Paragraphs aligned

How many paragraphs of the original were paired with a paragraph of the translation, shown as “N of M” on an open language. See The file header.

Partial support

Languages whose line breaking works differently: right-to-left, no spaces between words, or complex scripts. Everything is checked and counted; checks that depend on line breaks are a hint. See Languages.

Project

A job for a client: it holds the files, every analysis run on them and your review. See Projects.

Project analysis

A written overview of the whole project by the AI, on the Summary tab: what is most serious, what repeats, what is fine, what to tell DTP and the translator. It costs, and is generated only when you ask. See Ask the project.

QA comment

A comment a person drew on the page, for a problem EasyQLab didn’t flag. It counts like a finding and is signed with the author’s name. See Adding a QA comment.

Quick analysis

An analysis outside any project. Files are not stored, and the result can’t be reopened later. It is billed all the same. See Your first project.

R – S

Real error

The button that confirms a finding. It counts even if EasyQLab was unsure, and raises EasyQLab’s confidence in that kind of finding. In the Summary it is counted under Is an error. See Deciding.

Rejected

The red verdict: 3 or more critical findings, or 12 or more high ones, with the default thresholds. It can’t be delivered as is. See The traffic light.

Review complete

The button you click when you have gone through every finding of a language. It records who reviewed it and when. See Review complete.

Review profile

A saved configuration — which checks run, their severities, the traffic-light thresholds — that a project uses. Typically one per client. See Settings and review profiles.

Run

One analysis of a project. Runs are numbered: “Run number 2 of this project”. Each run keeps the settings it was made with. See Projects.

Second round

Running the corrected file of one language against the same original, to see what was resolved, what is still there and what is new. Only that language is billed. See Second round and human review.

Sensitivity

The setting that decides the confidence cut-off. It can be changed only on some plans, and a saved run keeps the one it ran with. See Sensitivity.

Severity

How serious a finding is, from 1 (acceptable) to 5 (critical). Each check has a default you can change. See Severities.

Side by side

The viewer mode with the original on the left and the translation on the right. See The viewer.

Summary

The first tab of the results: the project in one table, one row per language, with Copy summary, Download CSV and the Project analysis. See The Summary tab.

T – V

Target (Translation)

A translated file. On the upload screen, the Translations box takes one file per language. In a finding, “Target” is the evidence from the translation.

Threshold

One of the four numbers that turn severity counts into a verdict, such as “Rejected at this many critical (5)”. Set per review profile. See Changing the thresholds.

Traffic light

The three verdicts — Approved, Needs review, Rejected — in green, amber and red, for each language and for the project. See The traffic light.

Translation review

The AI review of the translation (C39): it reads each paragraph against its original. Its findings can be shown or hidden with Show translation review (N). See The translation review switch.

Verdict

Approved, Needs review or Rejected, always with its reason. Each language has one; the project takes the worst of its languages. See How the project gets its verdict.

Visual review

The AI visual review (C40): it looks at both pages side by side like a DTP reviewer. It needs the PDF. See What it checks.