Usage rates for MT have made steady gains in recent years it is near a tipping point: It is now on a path to become a mainstream solution for both enterprises and LSPs within the next three years.
Not only that, it’s a growing opportunity if we look in the context of language services as a whole
Right now, MT is a small portion of that.
We can’t say exactly how much, but…
What we can say is that it’s growing, and it’s only going to get bigger year after year…
BUT not only is it growing within traditional language services, it’s creating it’s own market outside of that, in cases where translation and MT was previously viable
Growing in key areas where previous there was not even an application because it wasn’t practical
We’re definitely now at a TIPPING POINT!
That’s all well and good, but which MT?
We’ve developed an MT framework that allows us to combine all types of MT, and natural language processing approaches, to chose the most appropriate solution for the task at hand.
We call this THE ENSEMBLE ARCHITECTURE
It does an On-the-fly combination of different translation outputs, from different technology – SMT, RULES, LANGUAGE SPECIFIC LIKE SPANISH, PHARMA, etc.
to ensure you get the best possible translation for a given language, content type or writing style
All while HIDING THE COMPLEXITY from the end user
One word that’s been conspicuous by it’s absence is “NEURAL”.
Hot topic, controversial at times with a lot of hype and research but LITTLE PRACTICAL CONSIDERATION
We’re changing that.
We decided to collaborate with Adapt to build neural MT and compare to state of the art large scale production engines – not just in terms of BLEU scores, but translation quality, and practicality.
Taking commercial state of the art and combining it with cutting edge academic research
KEY DIFFERENCE IS WE’RE DOING IT RIGHT
REAL WORLD NO SANITIZED DATA FIND OUT WHAT IT’S GOOD FOR
THERE’S VALUE TO BE ADDED, HOW CAN WE HARNESS?
We literally already have the perfect environment to allow NMT to be another string in the bow and let us use the most appropriate MT for the job
WHETHER IT BE NEURAL FOR KOREAN, FOR CHAT TEXT, OR WHATEVER THE CASE MAY BE
It’s not a one size fits all solution and who knows when it will be, but we have developed a framework that allows us to leverage it’s strength on a case by case basis to deliver the best possible translation for a given task.
Overtime we fully expect the “brain to grow” and become the best MT on offer for various language pairs and content types, and when it is, WE”RE PERFECTLY POSITIONS FROM A TECHINCOLOGY AND EXPERTISE PERSPECTIVE to capitalise on this wave.
Iconic Translation: The Neural Frontier by John Tinsley (Iconic Translation Machines)
The Neural Frontier
“Machine translation is on a path
to becoming a mainstream solution”
Chinese to English
Doing it right
Apples to apples
Access to same
training data, test
data, including all of
the ugly parts.
so what MT good and
what and where does
it fall down?
Iconic Neural MT
Neural MT works – and it’s good!
It is not a silver bullet
+ word order
- omitting phrases
+ error free output
- sentence structure