| Pages in topic: < [1 2 3] > | What do you think about this new predictive-typing translation technology? Thread poster: Jeff Whittaker
|
|---|
| Our memory may actually be better than we think | Feb 29, 2016 |
LegalTransform wrote:
it would be nice to get a reminder of how you previously translated a phrase internally within each document.
For example, when you start translating something like "la Ley del Organismo Judicial" or "Asociación para Evitar la Ceguera en México", the system would automatically remind you of how you translated it previously in that same document.
Once upon a time, I had to translate the same fragment of legal text (about 300 words) for two different clients. The jobs were almost a year apart, and the second time I didn't have access to my original translation, as I was on the road and the archive was on the external disk. However, upon returning home, I compared the two translations, and to my surprise, the difference between the two was negligibly small. On the other hand, donning my editor hat, I do occasionally see translators organically unable to translate several occurrences of the same expression identically, even when they are within the same job, on the same page! This software will certainly benefit them, but, well, call me a grumpy old man, but I consider these people professionally unfit to be translators.
| | | | Stepan Konev 
Russian Federation
Local time: 16:39
English to Russian
| It sounds like something that went out with the ark | Feb 29, 2016 |
Do you need to be able to retype and reproduce texts by heart to feel fit?
Any CAT tool can do it in a fraction of second. Without any 'negligibly small' difference at all. (The heading only refers to this first half of my post.)
What regards Lilt, I cannot understand why Lilt is better than other existing solutions (Autosuggest in Trados, Muses/Predictive typing in memoQ, or AutoWrite in DejaVu), but I will follow it to have a deeper insight.
[Edited at 2016-02-29 08:1... See more Do you need to be able to retype and reproduce texts by heart to feel fit?
Any CAT tool can do it in a fraction of second. Without any 'negligibly small' difference at all. (The heading only refers to this first half of my post.)
What regards Lilt, I cannot understand why Lilt is better than other existing solutions (Autosuggest in Trados, Muses/Predictive typing in memoQ, or AutoWrite in DejaVu), but I will follow it to have a deeper insight.
[Edited at 2016-02-29 08:10 GMT] ▲ Collapse
| | | | | A compromise between MT and offline predictive typing tools | Feb 29, 2016 |
I assume (although yet don´t know) that Lilt is just a technological compromise between the algorithms, on which Machine Translation and their still bad results are based, and predictice typing technologies like AutoWrite (Deja Vu), "Muse" (MemoQ) and Autosuggest (Studio), but online based. Lilt seems just to pick out the single-word or phrasal raisins from the meanwhile enormous cloud ressources of bilingual word exchange results (= translations) instead of assembling them into clumsy and faul... See more I assume (although yet don´t know) that Lilt is just a technological compromise between the algorithms, on which Machine Translation and their still bad results are based, and predictice typing technologies like AutoWrite (Deja Vu), "Muse" (MemoQ) and Autosuggest (Studio), but online based. Lilt seems just to pick out the single-word or phrasal raisins from the meanwhile enormous cloud ressources of bilingual word exchange results (= translations) instead of assembling them into clumsy and faulty segments à la Google Translate.
It may help to save typing effort but I never would use it to rely on specif terms this semi-intelligent solution proposes. And I would doubt that using Lilt doesn´t feed the cloud with your results too, when using it.
Compared with the major CATs, I still regard Deja Vu´s offline solution AutoWrite as the best predictive-typing technology: it´s offline, needs no extra tool to buy, no "Muses" to train, there is no minimum of segments in your TM or TB necessary and therefore works dynamically and on the fly.
[Edited at 2016-02-29 08:20 GMT] ▲ Collapse
| | | | Tom in London
United Kingdom
Local time: 14:39
Member (2008)
Italian to English
This b....from Lilt has started sending me really annoying unsolicited emails just because I clicked on something on his website.
Another reason not to have anything to do with Lilt.
If you're reading this, Lilt Boy: GO AWAY.
| | |
|
|
|
| At least it works! | Feb 29, 2016 |
Tom in London wrote: Lilt has started sending me really annoying unsolicited emails just because I clicked on something on his website.
Great! It seems they employ that tool even on their website: As soon as you hover over their web pages, Lilt automatically completes and fills in your data in the "Send me more info!" section. Next step: predicting your thoughts...
| | | | Tom in London
United Kingdom
Local time: 14:39
Member (2008)
Italian to English
Matthias Brombach wrote:
Next step: predicting your thoughts...
You can probably already predict my thoughts.
| | | | Sara Massons
France
Local time: 15:39
Member (2016)
English to French
+ ...
| Not specific to Lilt - automatic completion definitely not for me | Feb 29, 2016 |
I agree with most of you. I already tested this in other CAT tools and found it anoying and even driving my translation to the worst as I was tempted to leave the automatically proposed text without thinking before I read it. I happen to do some web dev as well and I've always been anoyed by the auto-completion tools in the editors.
Nevertheless, I once though I would never use a CAT tool at all (and neither an advanced Html editor or even a smart phone) and I finally get used to it... See more I agree with most of you. I already tested this in other CAT tools and found it anoying and even driving my translation to the worst as I was tempted to leave the automatically proposed text without thinking before I read it. I happen to do some web dev as well and I've always been anoyed by the auto-completion tools in the editors.
Nevertheless, I once though I would never use a CAT tool at all (and neither an advanced Html editor or even a smart phone) and I finally get used to it (except for the SMS auto suggesting !) so lets say wait and see what we do with that in a few years... ▲ Collapse
| | | | Spence Green
United States
Local time: 06:39
Member (2014)
English to Arabic
+ ...
| Emails from Lilt | Mar 24, 2016 |
Tom in London wrote:
This b....from Lilt has started sending me really annoying unsolicited emails just because I clicked on something on his website.
Another reason not to have anything to do with Lilt.
If you're reading this, Lilt Boy: GO AWAY.
Tom:
The only ways that you can receive emails from us are through your ProZ account or your Lilt account. If you don't want to receive third-party emails from ProZ, then you need to change your preferences. If you don't want to receive emails from us, then you need only click on the "Unsubscribe" link at the bottom of any email from Lilt.
I'm saddened that you found it easier and more productive to login to ProZ and post insults than to click an unsubscribe link.
Spence
| | |
|
|
|
Spence Green
United States
Local time: 06:39
Member (2014)
English to Arabic
+ ...
| Questions about Lilt / interactive MT | Mar 24, 2016 |
Matthias Brombach wrote:
I assume (although yet don´t know) that Lilt is just a technological compromise between the algorithms, on which Machine Translation and their still bad results are based, and predictice typing technologies like AutoWrite (Deja Vu), "Muse" (MemoQ) and Autosuggest (Studio), but online based. Lilt seems just to pick out the single-word or phrasal raisins from the meanwhile enormous cloud ressources of bilingual word exchange results (= translations) instead of assembling them into clumsy and faulty segments à la Google Translate.
It may help to save typing effort but I never would use it to rely on specif terms this semi-intelligent solution proposes. And I would doubt that using Lilt doesn´t feed the cloud with your results too, when using it.
Compared with the major CATs, I still regard Deja Vu´s offline solution AutoWrite as the best predictive-typing technology: it´s offline, needs no extra tool to buy, no "Muses" to train, there is no minimum of segments in your TM or TB necessary and therefore works dynamically and on the fly.
[Edited at 2016-02-29 08:20 GMT]
Matthias:
Re: "bad results", could you post an example or send me your Lilt username so that I can better understand the data for which the system failed? I can't find your name in our system....
The interactive MT technology behind Lilt is described in this paper: http://aclweb.org/anthology/D/D14/D14-1130.pdf
Re: "feed the cloud," I'm not sure what you mean, but I suspect that you're referring to third-party disclosure. We don't use third-party services like Google Translate. Your data is encrypted in our data center, which is covered by an enterprise-grade SLA: http://lilt.com/security
Re: "no training, no minimum segments," Lilt also satisfies these requirements. Learning happens in real-time while you work. No configuration---other than optional uploading of TMX files---is required.
Spence
| | | | | Tried it, liked it | Apr 6, 2016 |
I checked out lilt after seeing them here on Proz. Despite not being as feature rich as the offline tools that I use, I think it has a great deal of potential.
It was interesting to see how the results actually improved and became less MT-like as my TM grew. The suggested translation actually appears in a different field and you have to manually insert them with a click or hotkey combo; if you want to use them. Otherwise you type in your translation as with any other tool.
... See more I checked out lilt after seeing them here on Proz. Despite not being as feature rich as the offline tools that I use, I think it has a great deal of potential.
It was interesting to see how the results actually improved and became less MT-like as my TM grew. The suggested translation actually appears in a different field and you have to manually insert them with a click or hotkey combo; if you want to use them. Otherwise you type in your translation as with any other tool.
It's still missing features like auto propagation and joining/splitting segments for example, but I believe those are in the works.
I've communicated with the staff at severely odd hours and always got near immediate responses, which led me to wonder if they even sleep. They seem to appreciate feedback and suggestions, which gives me the impression that they want to create a viable product.
I think we should keep an open mind and see how it develops before attacking like a republican presidential candidate.
[Edited at 2016-04-06 07:53 GMT] ▲ Collapse
| | | | Tom in London
United Kingdom
Local time: 14:39
Member (2008)
Italian to English
Michael J.W. Beijer wrote:
... will it only work with Moses
I thought Moses died quite some time ago.
| | | | Tom in London
United Kingdom
Local time: 14:39
Member (2008)
Italian to English
Roy Williams wrote:
I checked out lilt after seeing them here on Proz. Despite not being as feature rich as the offline tools that I use, I think it has a great deal of potential.
It was interesting to see how the results actually improved and became less MT-like as my TM grew. The suggested translation actually appears in a different field and you have to manually insert them with a click or hotkey combo; if you want to use them. Otherwise you type in your translation as with any other tool.
It's still missing features like auto propagation and joining/splitting segments for example, but I believe those are in the works.
I've communicated with the staff at severely odd hours and always got near immediate responses, which led me to wonder if they even sleep. They seem to appreciate feedback and suggestions, which gives me the impression that they want to create a viable product.
I think we should keep an open mind and see how it develops before attacking like a republican presidential candidate.
[Edited at 2016-04-06 07:53 GMT]
Great piece of viral advertising there. I see this campaign is continuing unabated.
| | |
|
|
|
Michael Beijer
United Kingdom
Local time: 14:39
Member (2009)
Dutch to English
+ ...
Tom in London wrote: Roy Williams wrote:
I checked out lilt after seeing them here on Proz. Despite not being as feature rich as the offline tools that I use, I think it has a great deal of potential.
It was interesting to see how the results actually improved and became less MT-like as my TM grew. The suggested translation actually appears in a different field and you have to manually insert them with a click or hotkey combo; if you want to use them. Otherwise you type in your translation as with any other tool.
It's still missing features like auto propagation and joining/splitting segments for example, but I believe those are in the works.
I've communicated with the staff at severely odd hours and always got near immediate responses, which led me to wonder if they even sleep. They seem to appreciate feedback and suggestions, which gives me the impression that they want to create a viable product.
I think we should keep an open mind and see how it develops before attacking like a republican presidential candidate.
[Edited at 2016-04-06 07:53 GMT] Great piece of viral advertising there. I see this campaign is continuing unabated.
Great piece of viral anti-advertising there. I see your campaign is continuing unabated.
[Edited at 2016-04-06 08:14 GMT]
| | | | | You assume much | Apr 6, 2016 |
Tom in London wrote:
Great piece of viral advertising there. I see this campaign is continuing unabated.
So what exactly are you accusing me of Tom?
I simply voice my opinion based on my initial experiences with the tool. That is all.
Unlike you, I'm not as quick to scream witchcraft.
| | | | Michael Beijer
United Kingdom
Local time: 14:39
Member (2009)
Dutch to English
+ ...
I wouldn't really worry about Tom, he doesn't even use a CAT tool, so his opinion in such matters will obviously be, how shall I put this? ...slightly prehistoric.
I don't think Lilt will ever convince me until they transform it into a plugin so it can be used inside memoQ, but I concede that the idea is interesting. I tried it briefly, but just can't stand the online interface. Any online interface really. I want to work in my own CAT tool, with all my own resources.
... See more I wouldn't really worry about Tom, he doesn't even use a CAT tool, so his opinion in such matters will obviously be, how shall I put this? ...slightly prehistoric.
I don't think Lilt will ever convince me until they transform it into a plugin so it can be used inside memoQ, but I concede that the idea is interesting. I tried it briefly, but just can't stand the online interface. Any online interface really. I want to work in my own CAT tool, with all my own resources.
So Spence et al., if you are reading this: IMNSHO, the key to your success will be developing a plug-in for the major CAT tools.
Michael
[Edited at 2016-04-06 08:21 GMT] ▲ Collapse
| | | | | Pages in topic: < [1 2 3] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » What do you think about this new predictive-typing translation technology? | 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! » |
| | CafeTran Espresso | You've never met a CAT tool this clever!
Translate faster & easier, using a sophisticated CAT tool built by a translator / developer.
Accept jobs from clients who use Trados, MemoQ, Wordfast & major CAT tools.
Download and start using CafeTran Espresso -- for free
Buy now! » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
Translation news
