Types of LQA report – description and use case

Owned by Damian Rozmiarkowski (Unlicensed)
Last updated: Jul 11, 2024 by Damian Rozmiarkowski
6 min read

Introduction

There are two types of LQA report that can be generated in the XTM Cloud UI:

  • Project (general) LQA report;

  • Linguist (evaluative) LQA report.

It is crucial to understand the difference between these two reports because they are completely different from each other.

Project (general) LQA report evaluates a specific project as a whole.

Linguist (evaluative) LQA report evaluates the manual work done by a particular linguist.

Project (general) LQA report

Project (general) LQA report simply evaluates a specific project as a whole, and all the errors which have been added throughout the project (regardless of the workflow step in which they were placed) appear there. It is updated/generated up to date. The only prerequisite is that a particular CAT step has the LQA option enabled in the Workflow editor.

It can be found and downloaded from the Download section, selectable from of the project's context menu (Projects → Project list → (select a project and click on its context menu) → Download → LQA report):

You can also download this report for a specific file from its context menu (Project editor → Workflow → (select a file and click on its context menu) → Reports → LQA report):

Linguist (evaluative) LQA report

Linguist (evaluative) LQA report is for a particular linguist and is used to evaluate their work. Consequently, creating it is more complex as certain conditions have to be met.

LQA evaluation procedure

  1. First ensure that your workflow contains these two essential steps:

  • Translate/Correct/Review step (evaluated step),

  • LQA step (evaluating step).

IMPORTANT!

The evaluating step does not necessarily have to be an LQA step: it can be any CAT tool step in which LQA is performed. In addition, the option for saving the results in an LQA report must be enabled in the Workflow editor.

The Yes, don’t save result in user record enables the user to impose LQA errors but no evaluative LQA report will be generated from them. The only purpose of this option is to display potential errors at XTM Workbench level to other workflow participants (as long as the No, but show existing errors option is selected for a particular workflow step).

  1. Once the steps above have completed, relevant linguists need to be assigned to the appropriate steps:

  • evaluee (Translate/Correct step),

  • evaluator (LQA step).

  1. Evaluee enters the first step → performs ANY modification of target segments (whether that is actual translation or simply changing the segment status to "DONE") → confirms the segments ("green tick") → finishes the step.

  1. Evaluator enters the second step → performs LQA evaluation on the segments modified by the evaluee (imposes LQA errors on segments or does not) → confirms the segments ("green tick") → finishes the step.

  1. What you should end up with at this stage is the LQA report, which is stored in Project editor → LQA:

The way that this LQA type works is that it evaluates the linguist's work, which they performed manually in the preceding step. So the linguist must first do some manual translation in the first step and the LQA errors must then be imposed on the translated segments (and the segments later need to be confirmed) in the next LQA step. That is because the mechanism exclusively looks at the data from the reviewer's statistics in Project editor → Statistics → (LQA reviewer's statistics).

The total score in the LQA report from the screenshot above is calculated in accordance with the reviewer's word count which, in turn, is taken from Statistics.

The way that this LQA type works is that it evaluates the linguist's work, which they performed manually in the preceding step. So the linguist must first do some manual translation in the first step and the LQA errors must then be imposed on the translated segments (and the segments later need to be confirmed) in the next LQA step. That is because the mechanism exclusively looks at the data from the reviewer's statistics in Project editor → Statistics → (LQA reviewer's statistics).

The total score in the LQA report from the screenshot above is calculated in accordance with the reviewer's word count which, in turn, is taken from Statistics.

EXAMPLE

  1. Imagine a project workflow that contains two steps: Translate and LQA. In the Translate step, there are two segments.

  2. The linguist translates just one segment and leaves the second one unchanged because it is, for example, an ICE match that does not require modification. The linguist finishes the step.

  3. In the LQA step, the evaluator imposes two LQA errors, one on each segment, and finishes the step.

  4. In the LQA report, you will only see the LQA error that has been set for the segment that was modified by the linguist in the Translate step. This is because, as mentioned, the LQA feature only evaluates the linguist's work, not the segments that had been translated before by someone else.

  5. In a particular segment in the step before LQA, some change must be made by the linguist for the LQA error set for this segment in the LQA step to be visible in the report.

You can specify if the LQA step should be based on source or target statistics in Configuration → Settings → LQA → LQA options:

Where can these LQA reports be obtained?

  1. In Project editor → LQA:

  1. The user's Rating Tab (if acted as the "evaluee") and the Tasks Tab (if acted as the "evaluator"):

  1. If a linguist has the relevant LQA rights enabled in their Access rights:

They can see and download the reports from the LQA section:

  1. As for subcontractors, the LQA report will be listed/downloadable on the LSP side in Projects → LSP LQA, once the Contractor grants them the relevant access rights in the LSP's settings.

IMPORTANT!

An LSP cannot add the LQA step on their own. The Contractor first needs to add this step and assign the LSP to it.

LQA Simple & Extended forms

There are two versions of this LQA report type: simple and extended.

The only difference between them is that the extended version also contains the Extended table section displaying all the segments on which LQA errors were imposed.

LQA Per file & Per language reports

Also, there is a division into:

  • Per file → reports displaying LQA output limited to a specific file;

  • Per language → reports containing all the files (separately) within one document, sorted by target language.