KantanAnalytics generates quality estimation scores for automated translations generated by KantanMT engines. The quality estimation score is designed to be similar to a Fuzzy Match report from a TM. The more segments there are with a higher % quality score, the better the translation is. This means translations are more accurate and fluent and require less post-editing effort.
When a KantanAnalytics job is run, it creates: a translated source document, an XLIFF of that document and also a detailed project management report of all the segments in your translation. This includes segment-by-segment quality estimation scores in addition to other useful project statistics such as word count, character count, placeholder and tag counts.
KantanAnalytics can help Project Managers predict the cost and post-editing effort for Machine Translation projects. Prioritizing the right translations through segment quality estimation will yield the fastest possible project turn-around.
How to use KantanAnalytics
From the Dashboard, select the engine you wish to use. Go to the Training tab. Upload the file you want to analyse and click the Analyse button. A job will be launched and you can track its progress on the Jobs page.
When the job is finished, download the zip from the Jobs page. In the zip you will find:
- A translated file in the same format as the source file
- A bilingual XLIFF file which can be used in your CAT tool
- A KantanAnalytics quality report
Beside the Download icon, you will see a graph icon. Click on this to view your Analysis report. The report will show the percentage of the document that was leveraged from your Training Data - Total Recall - and also the quality score given to the rest of the document.
For more details on KantanAnalytics please see the following video below: