|
ADVERTISEMENT
|
|
British
scientists say: Tool Box Journal readers are smarter than the
rest. We have not checked it, but we have a bonus NY resolution for
you: Become more organized and productive this year. Work smart, not
hard, with AnyCount, TO3000, and Projetex!
Win
a discount from 25% to 90% with the Wheel of Fortune. Try your luck at translation3000.com/aitpn/557-78.html.
The
wheel stops on January 20 -- get it or regret it!
|
|
1.
I told you so!
|
|
I
know we've always tried not to say this to our kids, but after reading
the article I wrote in issue 297 of the Tool Box Journal
about ModernMT,
it's really hard not to at least think it. I described a refreshingly
different and capable machine translation system (I'll give a short
overview down below) with two major problems: the outrageously high
pricing and the completely outdated privacy concept that used any
customers' data for general training purposes. (Davide Caroselli,
ModernMT's VP of product, whom I talked to for both that article and
this, defended their privacy concept by saying: "You're right, we also
find it difficult at first to explain the privacy issue to our
customers.")
Well,
I'm pleased to announce that they have completely overhauled both their
pricing and their privacy considerations. There is no longer any
cross-training, and any data that you upload to enhance your own
machine translation will be strictly used by you only (see here). Kind of what
you'd expect from a paid product these days. (By the way, the data is
stored in their own data center in Italy, and they'll be adding a
larger data center soon in the US to accommodate AirBnB, their flagship
client.)
The
pricing is in line with other tools. You pay by the number of
characters as an LSP or translation buyer (between $8 and $50 per
million characters, depending on whether you want to train the engine
as you translate or queue documents for a batch translation) and a
monthly fee if you are a freelance translator ($25).
ModernMT
differs from its competitors (at least the ones off the shelf) because
you adapt it to your own data and then continue to adapt it without
actually spending time to train it. Its base engine is trained on the
large data sets collected by Translated, the Italian LSP and tech
developer of the massive MyMemory TM (Translated also owns a
majority of the shares in ModernMT). And while the baseline
engines are retrained once or twice a year, your data effectively sits
in the middle and adapts the MT suggestions to your style and
terminology. The system uses a technology called "instance-based
adaptive NMT," which sends translation requests to a TM layer
(consisting of even a relatively small TM as long as it's highly
tuned). Once similar segments are found in that TM layer, the NMT
engine's "hyperparameters" are adapted on-the-fly to generate a more
suitable suggestion. (The concept is based on this
paper by the Fondazio Bruno Kessler.)
I
asked Davide how many freelance translators are using the tool with the
old payment plan, and not surprisingly it's only a little handful. I've
talked to one who praised the translation quality as far superior to
other engines, a result that Intento has also concluded in one
of their reports (see here
-- note that they have also worked for ModernMT so there might
be a conflict of interest). Either way, if you are inclined to work
with adaptive MT, this might be worth a try. Presently you can use it
directly within SDL Studio Trados with this app, and of course in MateCat (owned by
translated) or via Intento (see elsewhere in this Journal).
|
|
ADVERTISEMENT
|
|
Speed
up Your Translation Processes with Across v7
The
new version of the Across Language Server and the Across
Translator Edition is now available! We have addressed numerous
subject areas in order to improve the user-friendliness, to reduce time
to market, and to enable new working styles.
Get
your Across Translator Edition v7
|
|
2.
TransTools
|
|
When
I (once again) decided to write about TransTools, I realized I had (once again) forgotten why I
had shied away from it previously. There is so much to say about it! In
fact, it's not easy to know where to start.
Maybe
we should start with its creator, Stanislav Okhvat, or as he might also
be called, Gyro Gearloose
(aka عبقرينو, Daniel Düsentrieb, Georg Gearløs, Géo
Trouvetou, Lang Ling Lung, Archimede Pitagorico, 자
이로기어루스,
Κύρος
Γρανάζης, Sriegas Bevarztis, Szaki
Dani, Oppfinnar-Jocke, Pelle Peloton, Professor Pardal, Diodak, Petter
Smart, or really in this case: Винт Разболтайло), the endearing
inventor in Disney's world (though Stanislav is less error-prone than
Винт).
Stanislav
used to be a technical translator, but a few years ago he switched to
focusing on tool development and consulting. He had good reasons for
the change. By that point he had already amassed a large number of
tools that were developed with translators, editors, and LSPs in mind
-- though I would venture to say they have the potential to be very
helpful for others as well.
Presently,
Stanislav offers a couple of single-purpose products (Term
Morphology Editor -- I wrote about that in edition 298 of the Tool
Box Journal -- and the Excel File Splitter for splitting Excel
files to distribute evenly among translators and later merge them
again) and two products with collections of utilities (little programs
with specific purposes) that mostly integrate into MS Word and
other MS Office programs: TransTools and TransTools+.
What might be a little confusing (and will be "fixed" at some point in
the future) is that TransTools+ is not necessarily the "pro"
edition of TransTools; in fact, they can both be run side by
side, complementing each other. The truth (about the "pro" notion) is
that the tools in TransTools+ can mostly also be found in TransTools,
though they are geared up to the next power level. It's just that not
all of TransTools' tools have made that switch. . . .
Confusing? I TOLD YOU!!
So,
let's simplify a bit.
Let's
start with TransTools+, a paid tool ($35,
but it has a generous trial period), which has only a handful of
utilities:
- Hide /
Unhide Text
- Multiple
Find and Replace
- Highlighting
Tool
- Document
Processing Tool
These
all show up in the ribbon bar of MS Word:
The
Hide / Unhide Text utility is a tool that is clearly
geared toward users of translation environment tools. Most translation
environment tools allow for an option to not translate hidden text in Word
documents, so you can prepare your documents to hide exactly what you
don't want to translate by applying the hidden feature. With Hide /
Unhide Text you can do that by
- searching
for already highlighted text,
- searching
fora certain type of content (textboxes, footnotes, etc.),
- selecting
from a preset list of exclusions (such as all HTML or XML tags, or all
time stamps in SRT files), or
- using
extremely sophisticated search features to locate the text (for more on
that, see below).
The
Highlighting Tool is similar (only, as the name
implies, it applies highlighting), but it's useful for a different
purpose. For instance, it's helpful for a project manager to batch
highlight certain things in a document before sending it back to the
translator or editor. Or, vice versa, for a translator or editor to
communicate questions to a PM or client. Any of the steps or series of
steps can be saved and applied to later files. Plus, with the Document
Processing Tool, they can even be applied to any number of Word
files (DOCX, DOC, RTF) simultaneously.
This
brings us to the Multiple Find and Replace tool (whose search
features are also part of the Hide and Highlighting
tools). As the name implies, it can run a number of search and replace
processes simultaneously and, as in the other tools, save that list of
processes for later re-use. This version of the search & replace
tool (in contrast to the one in the TransTools collection --
the one without the +) super-helpfully provides access to a huge range
of regular expressions beyond the ones offered in Word,
alongside descriptors that help you choose which one to use:
This
document provides a helpful overview of ways to use the tools, and
a preconfigured list of common problems after using OCR on a document
also serves as a good guide for how to put together a list:
I
don't know about you, but I spend a lot of time having to come up with
effective ways of searching and replacing data in documents, often in a
number of steps and/or over and over with similar processes. This tool
is just about as helpful as I can think of when it comes to that.
Let's
talk about TransTools. (continued below)
|
|
ADVERTISEMENT
|
|
Tool
Box Journal
subscribers save more on memoQ
Don't
you have your own memoQ translator pro license yet? memoQ
offers 30% off on all new translator pro licenses for Tool Box
Journal subscribers!
Read
our guide on the purchase process and apply code CAM_TBJ202030_PS
in the memoQ webshop. Offer expires at midnight January 23 in any time
zone.
|
|
2.
TransTools (continued from above)
|
|
Let's
talk about TransTools. You can see from its ribbon bar that
there are many more tools:
As
I mentioned above, the three main tools of TransTools+ (Hide
/ Unhide Text, Multiple Find and Replace, Highlighting
Tool) can be found among the offered utilities but with a slightly
different name and in a less powerful state (such as without all those
fancy regular expressions beyond the regular Word ones). But
there are plenty of other interesting tools that virtually everyone who
has ever worked in technical translation will immediately understand.
Here are some good examples:
- Document
Cleaner -- to
get rid of typical problems after OCR'ing a file or converting a PDF or
otherwise having to deal with lots of unnecessary codes (comparable to
the well-known CodeZapper tool)
- Find /
Replace Excessive Spaces -- as
the name says and including those before and after punctuation marks
- Unbreaker -- to
remove incorrect paragraph marks after copying and pasting text from a
PDF
- Dual
Language Assistant -- to
convert a document into a table by putting each paragraph of the source
language into a cell in the left column
- Manual
Localization -- to
automatically convert language-specific decimal separators, including
non-breaking spaces
- Document
Format Converter -- to
batch-convert any number of Word files of all formats (RTF,
DOCX, DOC) to any other Word format or PDF
- Quotation
Magic -- to
insert correct language-specific quotes
- Correctomatic -- to
automatically correct certain words (helpful when going between
different forms of English)
- What is
this Symbol --
gives you the ANSI code for the character(s) you highlight
- etc.,
etc.
TransTools
optionally also installs tools in PowerPoint and Excel:
Most
welcome in PowerPoint might be the Change Language tool
(see the graphic above), which changes the spellchecking language in PowerPoint.
I think we all can appreciate how helpful that is, knowing how tedious
it is otherwise.
In
Excel, the Glossary Search tool might be the
most helpful: it allows you to search multiple glossaries with an
independent program that is also installed at one time.
By
the way, you'll need to pay for some of the tools in TransTools
after a 45-day trial period, including Unbreaker, Correctomatic,
and Quotation Magic (these three will be ported to the TransTools+
version at the end of the month) as well as the Spellcheck Assistant
(a tool to build up and use custom dictionaries) and the Document
Format Converter. The Tag Cleaner and Unbreaker
tools in Excel are paid as well, and TransTools for Visio
and TransTools for AutoCAD that we have not talked about here
are paid features, too. The TransTools version with all the
paid features costs $25.
I
can't imagine that you have not seen at least a couple of processes in
this wealth of tools that you like, and chances are they might even be
among the free offerings. Of course, the drawback of program-specific
tools like these is that, well, they're program-specific (i.e.,
specific to programs like Word, Excel, etc.). What
differentiates these is that the utilities they offer really are meant
to prepare files that are to be translated or have been translated for
use elsewhere (either into a different tool, such as a translation
environment, or by a different person with a different function, such
as an editor), so their use is potentially much larger than just making
life easier in MS Word. And if there is anything to the Винт
Разболтайло concept, the usefulness and range of tools will only grow.
|
|
ADVERTISEMENT
|
|
Join
us on a special journey and kick-start your year by finding out what SDL
Trados can do for you, as we share our resources designed to help
you accelerate your career as a translation professional.
Learn
more: sdltrados.com/landing/we-are-sdl-trados.html.
|
|
3.
This is Intent(ed) to Clarify Things
|
|
I
recently did an interview with Konstantin Savenkov of Intento about what his tool
does and how or whether it can be useful to you and me:
Jost:
Can you give us a short overview of what your company offers?
Konstantin:
As of today, Intento helps large enterprises to procure and deploy the
best-fit machine translation (MT) across a wide range of enterprise
scenarios. Our main product is Enterprise MT Hub, which
provides a universal API [=application programming interface, i.e., the
interface that lets applications "talk" to each other] to almost every
MT on the market, making it easy to integrate multi-engine MT portfolio
with all software systems enterprises have onboard.
Basically,
we customize and evaluate 10-15 MT systems on customer data, build a
set of custom and stock MT engines and route requests to them to get
the best quality in every language pair and scenario. The scenarios
include traditional localization, customer support (tickets and chats),
website translation, corporate translation portals, and some others.
Each of them has a specific purpose for machine translation, and also
specific data which may be used to customize and evaluate the MT engine.
We
take the data, use our tools to clean it, customize MT engines and
select the best one for each customer scenario and language pair, along
with the expected ROI. Then in production, our integration platform is
used to deliver MT where it's useful, typically a handful of places
across the company.
J:
Is this something that is relevant for freelance translators? Do you
actually have freelance translators among your customers? Or is this
mainly for LSPs and translation buyers?
K:
One of the side-effects of our technology is that our plugins [see
below for a list] work with lots of MT engines. Another thing which we
found is useful to freelance translators is our general-purpose MT
routing, where we route requests to the engine which is best for this
language pair according
to our benchmark.
We
experiment a lot with different client segments, including freelance
translators. Freelance translators are an important part of the
ecosystem, so we are looking into how to make our tools affordable for
them rather than see them as a revenue stream.
J:
Can you give us a list of the different tool integrations that are
available?
K:
The full list of Machine Translation systems we support is here: Alibaba
(General and eCommerce engines), Amazon, Baidu, CloudTranslation,
DeepL, Google (Basic, Advanced and AutoML), GTCom,
IBM, Kakao, Naver, Microsoft
(including custom models), ModernMT, Naver, PROMT,
SAP, several SDL and Systran
systems, Tencent, Tilde, Yandex and Youdao.
We
have plugins for memoQ, SDL Trados, and Matecat.
Some other CAT tools use us on the backend as MT integration tool
(e.g., Smartcat). We also have XLIFF connectors to TMS systems
(such as XTM and Memsource), a Chrome
extension, plugins for Microsoft Office, and connectors to
general-purpose enterprise integration platforms, such as BMC
or Mulesoft.
J:
Can you tell us a little about your thoughts on using generic (stock)
engines vs. customized engines?
K:
We work a lot with customized NMT engines: Globalese, Google,
IBM, Microsoft, ModernMT, Systran,
Yandex. In most of the cases, customization with TM
and/or glossaries improves the outcome a lot.
It
takes resources to prepare the data, cook the engine and maintain it
over time, so it should be a careful decision based on the project
budget. For small projects, customization hassles may exceed the effort
to edit stock MT by much. However, we see quite a bit of raw MT cases
that would not be possible at all without the custom NMT.
J:
In a Twitter
conversation a few weeks ago, you were quoted with an anecdote from
a client who is using your product and who found out that unedited, raw
machine translation was more successful with their customers because
the customers felt that the more polished, post-edited translations
were sponsored and therefore not to be trusted. At face value, of
course, this sounds kind of shocking, but I went to a recording of the
talk where you mentioned this and it really was not as outrageous as it
first sounded. You were referring to Chinese-to-English translations of
product descriptions where the vast majority were raw machine
translation and only a few stood out as post-edited. It kind of makes
sense that customers see that as suspicious and don't trust those
descriptors as much. The two conclusions that I draw from this are that
a) the unedited machine translated segments must have been pretty bad
if it was so easy to see the difference, and b) this anecdote certainly
does not mean that overall raw machine translation sells better than
edited or translator-translated data. Would you agree?
K:
This anecdote is about the specific idea that the translation should be
evaluated for the specific purpose the MT is used. End-users are rarely
interested in the linguistic quality per se; they are looking for
something else: increased conversion, customer satisfaction, turnaround
time, reduced headcount. There may be some huge surprises like the one
in this anecdote. There may be solutions to the issues, e.g., in
another case with bad source content, it turned to be much better to
avoid the translation altogether, replacing it with text generation.
|
|
ADVERTISEMENT
|
|
Memsource
Editor for Mobile:
Translation in the Palm of Your Hand
Process
urgent client requests, assign translators and resources, monitor
project progress and even review and edit jobs wherever you are right
from your mobile. Try it now,
free!
|
|
The
Last Word on the Tool Box Journal
|
|
If
you would like to promote this journal by placing a link on your
website, I will in turn mention your website in a future edition of the
Tool Box Journal. Just paste the code you find here into
the HTML code of your webpage, and the little icon that is displayed on
that page with a link to my website will be displayed.
If
you are subscribed to this journal with
more than one email address, it would be great if you could unsubscribe
redundant addresses through the links Constant Contact offers below.
Should
you be interested in reprinting one of the articles in this journal for
promotional purposes, please contact me for information about pricing.
©
2020 International Writers' Group
|
|
