Lilt: poor prediction in En-Ru pair
Thread poster: Yakov Katsman
Yakov Katsman
Yakov Katsman  Identity Verified
United States
Local time: 04:35
Member (2016)
English to Russian
Dec 31, 2016

Tried Lilt for En-Ru translation using large TM from previous work in Wordfast.
First Lilt suggested segment translation usually is similar to Google Translate or other tools.
But when i make even small change/edit/correction in beginning of phrase the rest of suggested phrase became really bad compare to initial Lilt suggestion.
The more you edit the phrase the worse the rest of phrase getting.
Is it only in En-Ru pair? How this predictive feature works in other pairs? D
... See more
Tried Lilt for En-Ru translation using large TM from previous work in Wordfast.
First Lilt suggested segment translation usually is similar to Google Translate or other tools.
But when i make even small change/edit/correction in beginning of phrase the rest of suggested phrase became really bad compare to initial Lilt suggestion.
The more you edit the phrase the worse the rest of phrase getting.
Is it only in En-Ru pair? How this predictive feature works in other pairs? Does it improve the rest of suggestion while you editing the beginning?
Thanks for feedback
Yakov
Collapse


 
Spence Green
Spence Green  Identity Verified
United States
Local time: 01:35
Member (2014)
English to Arabic
+ ...
Debugging translation quality Jan 3, 2017

Yakov:

I've looked briefly into your account and can see why you might be observing a problem. To preserve your privacy, and the privacy of your data, I will send you an email separately.

For everyone else, here's a hint: translation quality is best when your TM data is similar to the text that you're translating. Massive, heterogeneous TMs of the sort commonly curated by veteran translators are better divided into coherent subsections for the purposes of MT training. H
... See more
Yakov:

I've looked briefly into your account and can see why you might be observing a problem. To preserve your privacy, and the privacy of your data, I will send you an email separately.

For everyone else, here's a hint: translation quality is best when your TM data is similar to the text that you're translating. Massive, heterogeneous TMs of the sort commonly curated by veteran translators are better divided into coherent subsections for the purposes of MT training. Here are some other tips:

https://lilt.com/kb/troubleshooting/training-lilt

Spence
Collapse


 


To report site rules violations or get help, contact a site moderator:


You can also contact site staff by submitting a support request »

Lilt: poor prediction in En-Ru pair







Wordfast Pro
Translation Memory Software for Any Platform

Exclusive discount for ProZ.com users! Save over 13% when purchasing Wordfast Pro through ProZ.com. Wordfast is the world's #1 provider of platform-independent Translation Memory software. Consistently ranked the most user-friendly and highest value

Buy now! »
Trados Studio 2022 Freelance
The leading translation software used by over 270,000 translators.

Designed with your feedback in mind, Trados Studio 2022 delivers an unrivalled, powerful desktop and cloud solution, empowering you to work in the most efficient and cost-effective way.

More info »