| Pages in topic: < [1 2 3 4] > | What's your take on TM Town? Thread poster: Joëlle Bouille
|
|---|
Tomás Cano Binder, CT wrote:
as long as their NDA and privacy agreements allow it, which is not my case
Of course not. We all do extremely secret and important work! Except for me, that is. Most of the jobs I do - including ones for which I signed a NDA - end up on the Internet, sometimes before I could manage to upload them on TM-Town.
But it's quite simple, the TMs from jobs/clients you signed an NDA for, shouldn't end up on TM-Town, at least not before they appear on the Internet.
What keeps sneaky translators from uploading a pile of stuff they have not translated, to capture new work in fields in which they have no experience at all? Isn't it easy to trick TM Town into believing that they master more areas than they actually do
You can do that, but before you upload TMs, you are required to state whether you translated them yourself, or not (and a few more things). Besides, you can cheat a client only once this way, so it's not very attractive to take the risk and the trouble to upload other people's jobs. And for huge public TMs like the DGT, Kevin has algorithms in place.
Cheers,
Hans
| | | |
Joëlle Bouille wrote:
Excellent question indeed...
I can affirm that. Brilliant question. I asked it when Kevin started his project. By the way, it's beginning to look like you have an agenda on this subject. I have an agenda too. And you're on it.
H.
| | | | ..... (X)
Local time: 22:03
Hi Tomás,
Thanks for your question!
What keeps sneaky translators from uploading a pile of stuff they have not translated, to capture new work in fields in which they have no experience at all? Isn't it easy to trick TM Town into believing that they master more areas than they actually do? What defence would honest translators have in TM Town against these practices?
A few points. First, when a translator loads a document into their TM-Town account she/he needs to declare that document as one of the following:
- I translated this document myself
- I partially translated this document
- I only proofread this document
- This document is for reference only
TM-Town's system also analyzes documents for certain heuristics and red flags to detect public TMs and glossaries and may automatically classify a document as "Reference Only" and lock the classification. "Reference Only" documents are not used in matching and "Partially translated" or "Proofread" documents may only be used in matching in certain circumstances or given certain weights....or may not be used at all. We also keep a close eye on the site and are proactive about taking care of and checking out red flags (i.e. this person suddenly has 400,000 translation units but their profile says they have only been a translator for 6 months).
But in the end there is only so much we can do. It is very possible that a translator could load a document in which they do not have expertise in, just like it is very possible that a translator could state on their ProZ.com/Facebook/{any site} profile that she/he is, for example, a medical translator when in reality she/he is not. The goal therefore is to give site visitors (potential clients, LSPs, etc.) who are browsing the profiles as many tools and data points as we can so that she/he can form a judgment on that translator. This includes things like:
-Years of experience
-Education
-Associations or accreditations
-Links to other profiles (ProZ.com, personal home pages, etc.)
-Terminology Marketplace glossary sales - feedback and ratings from peer translators
So, Nakōdo search, just like our traditional directory search (or a ProZ.com directory search) returns a set of translators from which one can then drill down through to the translator's profile to get a better sense of whether that is the right person for the job. The difference with Nakōdo search being, instead of searching by some criteria you are searching by the text of the document.
Ultimately, the solution to a potential "cheating" issue is not merely a technological one, but a human one. Feedback from clients and peers. For example, Tomás, you have over two decades experience in high-tech translation and you have answered many KudoZ questions. If I were to ask you to recommend me an English to Spanish translator who knows industrial engineering...you could probably at least point me in the right direction. Or you would at least have some names that you would recognize as you have seen them around KudoZ or other places. This is an area that we intend to continue to pursue and incorporate more into TM-Town user profiles - feedback from clients and peers. This is where I think ProZ.com really does an excellent job with things like WWA and Certified ProZ that we hope to emulate, build on, or possibly even query through the ProZ.com API. Again with the goal of giving the client browsing translator profiles on TM-Town as many solid data points as we can so that she/he can quickly and easily find the best translator for the job.
Kevin
| | | | | Thanks a lot Kevin! | Dec 29, 2015 |
Thank you for the detailed explanation. It is reassuring to see that you do checks as well and customers have the option to build themselves an opinion about potential candidates for their work. I just hope TM Town does not get big too quickly, which would reduce your chances of doing the detailed checks.
| | |
|
|
|
Joëlle Bouille 
France
Local time: 15:03
English to French
+ ...
TOPIC STARTER
My only agenda is I don't like being lied to and I don't like being exploited.
And as for these 2 websites I'm working on, they have absolutely nothing to do with translation. No "competition" agenda here.
Save yourself the trouble on doing the research on me: it's human-path.org and not-for-cheaters.com (not online yet and won't be before a year).
Both are psychology related, from very different angles, because I would like to contribute a tiny bit to making the ... See more My only agenda is I don't like being lied to and I don't like being exploited.
And as for these 2 websites I'm working on, they have absolutely nothing to do with translation. No "competition" agenda here.
Save yourself the trouble on doing the research on me: it's human-path.org and not-for-cheaters.com (not online yet and won't be before a year).
Both are psychology related, from very different angles, because I would like to contribute a tiny bit to making the world a better place.
Nothing very threatening, I think. Do you? ▲ Collapse
| | | | | Before or after, I do not care | Dec 29, 2015 |
Meta Arkadia wrote: Tomás Cano Binder, CT wrote:
as long as their NDA and privacy agreements allow it, which is not my case Of course not. We all do extremely secret and important work! Except for me, that is. Most of the jobs I do - including ones for which I signed a NDA - end up on the Internet, sometimes before I could manage to upload them on TM-Town. But it's quite simple, the TMs from jobs/clients you signed an NDA for, shouldn't end up on TM-Town, at least not before they appear on the Internet.
It is not my prerogative to decide whether something is already public domain so that I can publish it wherever I like. I believe in the word given and contracts signed. I would not sign an agreement in which I do not believe or am not ready to honour. If I have committed never to post, upload, etc. any information from my customers or my translations for them, I believe in honouring such commitment, both for obviously private stuff and for obviously public stuff.
| | | | | No guarantee... | Dec 29, 2015 |
that my TMs wouldn't be exploited in any way. I'm not calling this a "scam", but just people telling me "we won't do it" is not enough for me. There's just too much at stake.
It would be interesting to know how successful the project has been until now...
| | | | Tom in London
United Kingdom
Local time: 14:03
Member (2008)
Italian to English
Meta Arkadia wrote: Joëlle Bouille wrote:
Excellent question indeed... I can affirm that. Brilliant question. I asked it when Kevin started his project. By the way, it's beginning to look like you have an agenda on this subject. I have an agenda too. And you're on it. H.
Me too !
| | |
|
|
|
| TM Town no different from the Web in this respect | Dec 29, 2015 |
Tomás Cano Binder, CT wrote:
What keeps sneaky translators from uploading a pile of stuff they have not translated, to capture new work in fields in which they have no experience at all ? Isn't it easy to trick TM Town into believing that they master more areas than they actually do? What defence would honest translators have in TM Town against these practices?
If that is what a translator wants to do, then they don't need TM Town to do it. Anyone here on Proz could upload samples of work that are not their own. Candidates for exams to professional translation organisations which test candidates via remote testing (candidates are sent a test which they have X time to complete and return) could call up a friend and have that friend sit the exam.
But it's not going to get you far in the real world, when someone purporting to specialise in a given area can't provide a decent translation when it comes to delivery time.
| | | | | Why TM-Town can make a difference | Dec 30, 2015 |
Fiona Grace Peterson wrote:
Anyone here on Proz could upload samples of work that are not their own.
True enough, but TM-Town (hereinafter TM-T) provides the samples, so a cheater will either have to steal a translation that's already on TM-T, or he'll have to ask the brilliant translator to translate it for him. The first is pretty easy to check/prevent by Kevin, the latter is very unlikely.
However, uploading sample translations is only "second best," it's a substitute only recently made available to people who don't want or can't upload TMs, for whatever reason. And uploading TMs does have many advantages:
You have a copy in the cloud
You can search one/more than one/all TMs simultaneously
And you can search them blistering fast, without having to use your own hardware resources
Uploaded segments count for your TM-T ranking, improving your "exposure"
And most importantly (we all hope), outsourcers will use your TMs to find you. They can upload the document (or parts of it, or keywords only) that needs to be translated, and TM-T's algorithm will find the most suitable candidate.
The results are shown without mentioning the search terms, and to the outsourcer only:
You can try it yourself here. By the way, it'll only show results for translators who opted-in.
TM-T offers heaps of other useful features, including OCR and aligning, too much to mention, really.
Cheers,
Hans
| | | | Joëlle Bouille
France
Local time: 15:03
English to French
+ ...
TOPIC STARTER | TM Town’s core claim: automated quality assessment | Dec 30, 2015 |
I’m still puzzled.
So TM Town’s core claim is finding the best translator for any given source text (whatever its alphabet or syntactic/semantical/morphological specificities) thanks to the patent pending Nakōdo search technology. In other words, finding the translator that will produce the best target text (again whatever its alphabet or syntactic/semantical/morphological specificities).
This requires (i) matching keywords in the client’s source te... See more I’m still puzzled.
So TM Town’s core claim is finding the best translator for any given source text (whatever its alphabet or syntactic/semantical/morphological specificities) thanks to the patent pending Nakōdo search technology. In other words, finding the translator that will produce the best target text (again whatever its alphabet or syntactic/semantical/morphological specificities).
This requires (i) matching keywords in the client’s source text with keywords in uploaded source material (nothing revolutionary here, whatever the alphabet of the source language: it’s just a matter of comparing text strings, with no syntactic/semantical/morphological considerations) but more importantly (ii) assessing quality of uploaded target material and/or translated TM Town samples .
What puzzles me is that assessing target quality seems far more complicated to automate through algorithms. It is, besides, even more important if translators are to be assessed and selected (in other words “ranked”) based on their rendering for “standard” TM Town provided samples.
Assessing target quality is at least twofold: (i) does target correctly render all ideas and nuances contained in the source and (ii) does it read like a properly written native text.
When considering what a properly written native text is, a few key terms may need reminding (taken from The Psychology of Language: from Data to Theory (see https://books.google.fr/books?id=Ip5cAgAAQBAJ&pg=PA5&lpg=PA5&dq="language%20morphology"%20definition&source=bl&ots=ornjC-7FXC&sig=baH-0-jDccy1FbLqHSLfZ2juNoQ&hl=en&sa=X&ved=0ahUKEwjTuujigYPKAhWGqA4KHX63AYEQ6AEIXDAK#v=onepage&q="language%20morphology"%20definition&f=false):
Semantics: the study of meaning
Syntax: the rules of word order in a language
Morphology: the study of how words are built up from morphemes
Morpheme: the smallest unit of meaning (e.g. “dogs” contains two, dog + plural s)
Inflectional morphology: the study of inflections (English virtually has none)
Pragmatics: aspects of meaning that do not affect the literal truth of what is being said; these concern things such as choice from words with the same meaning, implications in conversation and maintaining coherence in conversation.
All above elements are easily and quickly checked by a native human brain. But does the Nakōdo search technology also check these are correct? For any source or target language? Whatever their respective alphabets and syntactic/semantical/morphological specificities?
English syntax is very simple and its inflectional morphology is very poor compared to many languages, including French, Spanish, German, Italian, Russian or Arabic, to name only a few. English also pays much less attention to anaphoras, for example, compared to French or Spanish (to name languages I know), which has an impact on the internal consistency of a text and on its proper flow (this involves checking links are properly, i.e. natively, made between translation units and has to do with pragmatics).
By and large, the volume of language corpora available for English far exceeds language corpora available in other languages, because English is the international language of reference.
I’m no NLP expert but my feeling is that most NLP research covers English and is, as a consequence, conceptually limited, constrained, by the very “simplicity” and flexibility of the English language itself, because it assumes similar simplicity and flexibility for other languages.
But an algorithm that checks Russian semantics, syntax and inflectional morphology will look very different from one that does the same for English. And it will be more complex.
So I’m wondering. Does TM Town actually have specific algorithms in place for each and every language in the world (including logographic languages like Chinese, Japanese or Korean; tonal languages like Vietnamese, where proper tone, hence proper meaning, may only be implied from the context; or inflectional languages like French, Russian, Arabic or Japanese)?
And does each of these language-specific algorithms check for proper native syntax, semantics, morphology and pragmatics?
Considering TM Town provided samples are supposed to prevent “cheating”, such quality assessment becomes of the essence to discriminate between, say, 20 translators who chose the same sample.
But if all TM Town does is matching keywords in each source language and then actually leaves quality assessment to the client, in other words lets the client himself assess who is “the best translator”, I don’t see where the innovation lies. And if quality assessment is done by other translators, well, Proz does it already with the Proz.com certification program.
Any Project Manager already selects translators according to their specialty areas (that’s pretty much keyword matching).
Any serious translation agency already has very specifically and carefully chosen test translations, and has professional natives available to check for proper native syntax, semantics, morphology and pragmatics.
Unless TM Town truly automates quality assessment and performs it to a level that is comparable to what a native professional PM will achieve, it will just copy what Proz.com already does, and does very well, I think.
And as for TM management, confidentiality requirements dictate not to do it in the cloud.
[Edited at 2015-12-30 08:54 GMT]
[Edited at 2015-12-30 08:56 GMT] ▲ Collapse
| | | | ..... (X)
Local time: 22:03
| Standing on the shoulders of giants | Dec 30, 2015 |
You’re right, ProZ.com is a great platform that we have a lot of respect for. As Henry mentioned earlier in this thread we have talked extensively and he has provided a lot of great advice and mentorship.
In its current state maybe some of the things we are doing are only incremental improvements to the status quo, but I think the possibilities unlocked by being able to fully leverage your prior work as a translator and have that work working for you to attract you new clients in ... See more You’re right, ProZ.com is a great platform that we have a lot of respect for. As Henry mentioned earlier in this thread we have talked extensively and he has provided a lot of great advice and mentorship.
In its current state maybe some of the things we are doing are only incremental improvements to the status quo, but I think the possibilities unlocked by being able to fully leverage your prior work as a translator and have that work working for you to attract you new clients in your field(s) of expertise is the future. ▲ Collapse
| | |
|
|
|
Joëlle Bouille
France
Local time: 15:03
English to French
+ ...
TOPIC STARTER | The future of translation is for "translators" to understand they are linguists | Jan 3, 2016 |
Doesn’t any translator’s work already “work for her/him”?
Speaking of giants, I think future will be the end of giants making huge money on those who do the actual work, on “the masses”.
I think future will be discernment and empowerment of “the masses”, thanks to universal access to knowledge – reliable and meaningful knowledge.
I think future will be the disappearance of Facebookish vacuous content and marketing-like communication wrappe... See more Doesn’t any translator’s work already “work for her/him”?
Speaking of giants, I think future will be the end of giants making huge money on those who do the actual work, on “the masses”.
I think future will be discernment and empowerment of “the masses”, thanks to universal access to knowledge – reliable and meaningful knowledge.
I think future will be the disappearance of Facebookish vacuous content and marketing-like communication wrapped in sensationalism and “attractive” and colorful interfaces.
I think the future of translation will bring an increasing divide between quality-oriented LINGUIST-like translation on the one hand (i.e. the “writer’s approach” to translation), and automation-oriented machine-like “hard discount” translation on the other (i.e. the “software developer/programmer’s approach” to translation). This divide is already there (and it certainly doesn’t favor machine-translation or PEMT, considering overall quality of written materials keeps decreasing).
And once we reach the absolute bottom of the writing pit, I think the flaws of automated approaches will become obvious for everyone, and will promote a return to “old-school” quality-oriented approaches, which exclude any PM handling any language she/he knows nothing about (a very very very widespread practice nowadays, that TM Town’s “automated keyword matching” might really promote).
I guess it will take a few costly and resounding mistakes to get there.
In the meantime, I wish all linguists of the world a very prosperous and quality-oriented new year 2016!
[Edited at 2016-01-03 17:44 GMT] ▲ Collapse
| | | | | Road maintenance and universal knowledge | Jan 3, 2016 |
Joëlle Bouille wrote:
I think future will be discernment and empowerment of “the masses”, thanks to universal access to knowledge – reliable and meaningful knowledge.
I slightly disagree in this matter. Most people living in small, distant towns tend to think that they need better roads to guarantee the survival of their little settlement in the middle of population concentration. The fact is, however, that the better the roads to these places, the less people stay there. The reason is simple: it is more comfortable to live in a bigger place, and after all, you can always spend the weekend in your little town. Gradually, people tend to spend more and more weekends in the city, and the little town is ultimately abandoned... thus making the big road investment completely usless.
This looks like completely unrelated to your statement, but it is not: the easier and more available the access to information, the less information we try to remember. Internet is concentrating specialised knowledge, and the number of people who really know how to do something (from woodwork to accounting to computing to spelling) is shrinking by the minute. It is absolutely vital that we know our stuff by heart, without relying on online sources many times a day.
| | | | Joëlle Bouille
France
Local time: 15:03
English to French
+ ...
TOPIC STARTER
I totally agree with you, Tomás.
What I actually mean is that technology and information availability (and any automated process) make a lot of people intellectually lazy. This is the current first phase.
Current volumes of available information give the impression -- to too many people -- that we don’t need to “know” anything anymore and don’t need to take the time to truly understand anything anymore. But the truth is, HIGH QUALITY and MEANINGFUL information ... See more I totally agree with you, Tomás.
What I actually mean is that technology and information availability (and any automated process) make a lot of people intellectually lazy. This is the current first phase.
Current volumes of available information give the impression -- to too many people -- that we don’t need to “know” anything anymore and don’t need to take the time to truly understand anything anymore. But the truth is, HIGH QUALITY and MEANINGFUL information is still very scarce.
This is why overall quality of written materials keeps decreasing: because currently available information is all about quantity, not quality, and because it makes a lot of people intellectually lazy. Copy-paste rules!
What I think is that at some point (this may take some time), universal access to knowledge will make a majority of people more discerning and smart.
Which means a majority of people will see, for example, what it takes to produce a high quality translation. Or, to quote you, “that it is absolutely vital that we know our stuff by heart, without relying on online sources many times a day”, to produce any high quality, discerning, truly worthy work.
What I mean is that sooner or later, true knowledge and quality will win out over automated processes and intellectual laziness.
[Edited at 2016-01-03 14:57 GMT] ▲ Collapse
| | | | | Pages in topic: < [1 2 3 4] > | To report site rules violations or get help, contact a site moderator: You can also contact site staff by submitting a support request » What's your take on TM Town? | Anycount & Translation Office 3000 | Translation Office 3000
Translation Office 3000 is an advanced accounting tool for freelance translators and small agencies. TO3000 easily and seamlessly integrates with the business life of professional freelance translators.
More info » |
| | TM-Town | Manage your TMs and Terms ... and boost your translation business
Are you ready for something fresh in the industry? TM-Town is a unique new site for you -- the freelance translator -- to store, manage and share translation memories (TMs) and glossaries...and potentially meet new clients on the basis of your prior work.
More info » |
|
| | | | X Sign in to your ProZ.com account... | | | | | |
Login to reply/comment 
