Difference between revisions of "Quality score formula"

From TQAuditor Wiki
Jump to: navigation, search
(Quality score calculation examples)
(Quality score formula visualisation and interpretation)
Line 18: Line 18:
 
:In the end, the result is always converted to 1000, a most optimal value.
 
:In the end, the result is always converted to 1000, a most optimal value.
 
The mistake severity scores, mistake type weights, and reports are based on this 1000-word amount. For example, a critical mistake with score of 20 with weight coefficient of 1 reduces the quality score by 20 per 1000 words (by 10 per 2000 words, accordingly, and so on).
 
The mistake severity scores, mistake type weights, and reports are based on this 1000-word amount. For example, a critical mistake with score of 20 with weight coefficient of 1 reduces the quality score by 20 per 1000 words (by 10 per 2000 words, accordingly, and so on).
 +
 +
For more information, please check the [[Quality score formula:Details|Quality score formula:Details]] page.
  
 
=='''Quality score calculation examples'''==
 
=='''Quality score calculation examples'''==

Revision as of 13:51, 27 January 2022

Quality score formula visualisation and interpretation

QSF.png


1. Account score limit
The highest possible score (100 by default). Can be set on this page: [Evaluation settings]
2. Mistake severity score
Score of mistake by its severity (for example, 1 point for minor and 5 points for major mistakes). Can be set on this page: [Mistake severities list]
3. Mistake type weight
Weight of a specific mistake type by specialization. Can be adjusted on this page [Mistake types list]
4. SUM
A sum of products of #2 and #3 for all the mistakes.
5. Project evaluation word count
Automatic word count: The value from the "Total source words" field. Please see the Automatic vs. manual word count page.
Manual word count: The value from the "Evaluated source words" field (specified when starting the evaluation with a manual word count).
6. 1000
In the end, the result is always converted to 1000, a most optimal value.

The mistake severity scores, mistake type weights, and reports are based on this 1000-word amount. For example, a critical mistake with score of 20 with weight coefficient of 1 reduces the quality score by 20 per 1000 words (by 10 per 2000 words, accordingly, and so on).

For more information, please check the Quality score formula:Details page.

Quality score calculation examples

Automatic word count

Manual word count

TQAuditor v3.0

How Quality Score is calculated

In TQAuditor v3.0 the Quality Score formula is similar to MQM score formula http://www.qt21.eu/mqm-definition/definition-2015-12-30.html#scoring-algorithm

The quality score reflects the number of mistakes made per 1000 words of the translated text.


The formula is:

ROUND(GREATEST(account.score_limit - SUM(mistake_severity.score * mistake_type_spec.weight) / project.evaluation_total_word_count * 1000, 0), 2)


The formula interpretation:

ROUND(var1, 2) — returns rounded to two decimal places value

ROUND(GREATEST(var2, 0), 2) — returns "0" in case of a negative value (example: if "-5" then "0").


Variables interpretation:

var1 = GREATEST(main2, 0)

var2 = account.score_limit - (SUM(mistake_severity.score * mistake_type_spec.weight) / project.evaluation_total_word_count) * 1000

  • mistake_severity.score—the mistake severity score.
  • mistake_type_spec.weight—the weight coefficient of the mistake type per project specialization.


Note: account.score_limit is equal to 100 by default, but you may define the highest score limit you need in the Evaluation settings.

TQAuditor v2.14

How Quality Score is calculated

Quality score reflects the number of mistakes made per 1000 words of translated text.

The formula is:

Quality score = Σ(mistake_severity.score * mistake_type_spec.weight) * project.evaluation_corrected_word_count / project.evaluation_sample_word_count / project.evaluation_total_word_count * 1000, where:

  • Quality score - the number of "base" mistakes per 1000 words ("base" mistake - the mistake that has the severity score of 1 and the weight coefficient of 1).
  • Σ - the sum of products of the mistake severity score and the weight coefficient of the mistake type per specialization.
  • mistake_severity.score - the mistake severity score.
  • mistake_type_spec.weight - the weight coefficient of the mistake type per project specialization.
  • project.evaluation_corrected_word_count - source words in corrected units in the evaluation sample.
  • project.evaluation_sample_word_count - total source words in the evaluation sample.
  • project.evaluation_total_word_count - total source words after evaluation start filter is applied.

How the number of translated words is selected

To make it simpler, let’s make an example:

Before starting the evaluation, the reviewer selects the number of words to evaluate:

Start evaluation filter.jpg

Now imagine that the whole translated text (marketing specialization) contains 3126 words, and the editor corrects the segments containing 2323 source words.

You select 1000 words sample which is randomly taken by the system out of 2323 source words in corrected units and add 5 mistakes of different types and severities. Two of them are minor punctuation mistakes

(the severity score of 1 and the weight coefficient of 0.9 for every mistake), and three of them are major grammar mistakes (the severity score of 5 and the weight coefficient of 1.2 for every mistake):

QS types and severities 1.jpg

Note: By default, the system has pre-defined quality standards, but you can define your own corporate quality standards.

As a result, we get:

Quality score = Σ(1*0.9+1*0.9+5*1,2+5*1,2+5*1,2)*2323/3126/1001*1000=14,699

You’ll be able to see these numbers on the evaluation page:

QSF result 1.jpg

Where:

  • Quality score = 14,7.
  • project.evaluation_corrected_word_count = 2323.
  • project.evaluation_sample_word_count = 3126.
  • project.evaluation_total_word_count = 1001.

Why is it made that way?

We came to this through several stages of evolution.

First, the system was selecting just the beginning of the text, and we found out that the translators started to translate first 1000 words better than the rest of the text. So we decided to select the random part of the text. But then the system sometimes selected the pieces containing no corrections while skipping heavily corrected parts. So we changed the logic, and now the system returns the required number of corrections to the evaluator, but remembers how much text it took to find these corrections.