Hi, my name is Stephen Davis from Iconic Translation
Machines and today I wanted to talk to you briefly about how Neural Machine Translation
works in the Financial Services Industry, specifically in compliance.
Regulators want banks, brokers, asset managers, let’s call them Institutions, to maintain
records of all the activities they undertake. And some of those interactions are written,
some are voice, but the vast majority are digital.
Compliance teams within the institutions need to be able to supervise their colleagues’
activity but often the activity is in a foreign language, for instance a London based compliance
team monitoring its French traders. That’s where Neural Machine Translation
can quickly and effectively translate into the native tongue of the compliance officers
in order to gain an understanding of what was done, said or implied.
In the traditional ‘Three Lines of Defense model’ used by many institutions, what we
do is essentially a function of the first line, where automation is used to quickly
and efficiently wade through a huge amount of data.
Some of the text we are asked to translate has already been converted from audio to text,
so we are a secondary process which puts greater emphasis on our accuracy, or usability, as
what we have been asked to process might not be 100% accurate to begin with, though voice
to text transcriptions are getting a lot better. Much of our work in this area is within the
Relativity TRACE product. This is important as it maintains the workflow
for escalation to the second line of defense and also maintains a defensible chain of custody
of all the interactions that might be subjected to scrutiny by the compliance teams or the
regulator. It is a newish departure for us, but one that
is growing all of the time. If you want to know more about our involvement with this,
please contact us via the website and we’ll gladly explain more. Thank you for listening.