Skip to main content

Review/QC using MQM tools

Plint Subtitler supports a review mode where each change/error is logged according to a standardised set of issue types. Using these tools involves a combination of tools and settings rather than a single Delivery method:

MQM (Multidimensional Quality Metrics) is a framework for objective assessment of translation quality, using standardized error categories and severity levels to calculate a quality score. Depending on the demands, categories can be adapted/omitted and the threshold levels for acceptance adjusted.

Adjusting and understanding the MQM settings in Plint

Under Tools & Data / MQM Settings, a site admin specifies how the various error categories will affect the report. Each reported issue will add a number of points to the accumulated error score. This number depends on what value is entered here for the selected Category and Severity (e.g. "Minor"), multiplied with the Weight. In the below example, a Major Mistranslation would result in 120 points.  

The "Max allowed errors" setting determines if a job will Pass or Fail, based on the following calculation: is the combined score per 1000 words higher or lower than this threshold? This calculation is done when the report is completed and summarised (see below). 

It is highly recommended that you make any such changes early on, to ensure that reports aggregated over time can be compared in a meaningful way.

mqm_settings.png 

Plint Subtitler in MQM mode

The actual review is done in Plint Subtitler, in a dedicated MQM mode. How to use this mode is explained here. The mode is activated on the Editor settings tab:

mqm_review_mode.png

An MQM Report job (optional)

This job consists of viewing the report after the quality control has been completed. All reports are stored and linked to the users (Reporter / Translator) and can be accessed as needed for the users in question. However, in order to quickly assess the result of a test, you can create a job and assign it to a project manager, vendor manager or similar.

You create such a job with the Delivery method "MQM Report". 

The job itself will appear on the user's Dashboard with an icon leading to the report (red or green, depending on Pass/Fail status). There is also a button to report the job as Done and remove it from the Dashboard. Please note that this will not remove the report from the system:

mqm_report_jobs_on_dashboard.png

Reading the report

The report will summarise the result by showing the number of individual errors and their categories and severity, along with charts and statistics. You can also export the complete list of errors in Excel format. 

mqm_report.png

Aggregated data on user level

By going to a user's summary page, you can find all their MQM reports. They can be viewed individually, or summarised with the "Aggregated MQM Scorecard" link. 

mqm_user_results.png