Tony O’Dowd joins experts at the 14th biennial AMTA Conference on October 6-9, 2020 to share proven strategies for handling Tags, Placeholders and Terms in Neural Machine Translation. This presentation will take place on October 9th at 1pm (UTC-7).

22nd September, 2020: As use cases for Neural Machine Translation (NMT) further expand, it is now becoming clear that a focussed effort on streamlining NMT workflows is needed to meet rising expectations and drive increased adoption of NMT across both the localisation and commercial sectors.

Neural Machine Translation has been proven to outperform state-of-the-art Statistical MT systems by a factor of 6 or more, while delivering translation at scale and rapid speed, yet there is still great opportunity to increase the efficiency of NMT workflows in order to both simplify NMT management and to allow for more seamless integrations. It is also imperative for improving Neural Machine Translation quality, consistency and coherency.

During his session, Lexically Constrained Decoding for Sequence Generation, Tony O’Dowd, Chief Architect at KantanMT will discuss NMT pre-processing strategies that the team at KantanMT have tested on their own systems along with the results of a number of those strategies. The session will also offer best practices on introducing these strategies into NMT workflows.

“Through the development of our own NMT innovations and listening and working with client needs, we have gained important insight into how to best manage Tags, Placeholders and Terms in NMT,” says Tony O’Dowd. “I am pleased to share these insights with peers at the AMTA conference and look forward to hearing from other speakers. As a sector there is much to be learned from each other and I am pleased that AMTA gives us the forum to do that.”

KantanMT is a pioneer of Neural Machine Translation and successfully commercialised its innovation on the KantanMT Platform. Recently, KantanMT has launched a crowdsourcing translation environment, KantanSkynet, combining the latest Neural Machine Translation technologies with AI, and a community of editors, to deliver always-on, scalable, translation services.

Tony’s talk offers insights for researchers and localisation managers in improving their NMT workflow and highlights how NMT can be streamlined effectively to drive greater results and to broaden the application opportunities of Neural Machine Translation.

For those who cannot make the conference but would still like to benefit from the session, KantantMT will be holding a special live webinar on the topic on Thursday 15th October. Register to attend the webinar.

If you would like to discuss NMT strategies with the team at KantanMT. Please contact us. We’re always here to help.

About AMTA Conference:

The 14th biennial conference of the Association for Machine Translation in the Americas (AMTA 2020) will take place from 8:00 AM PDT October 6 through 3:00 PM PDT October 9, 2020 as a virtual conference. AMTA 2020 provides a unique opportunity for academic and commercial researchers to share their results with colleagues as well as to understand real-world user requirements. Business and government participants benefit from updates on leading-edge R&D in MT as they present and discuss their use cases.

About KantanMT

KantanMT is a pioneer in Neural Machine Translation solutions. Its market leading platform, KantanMT, enables global organisations to customise, improve and deploy neural MT solutions at scale. Their KantanSkynet platform enables modern enterprises to deliver digital content in native languages, with always-on, AI-enhanced, crowd-sourced translation services. KantanSkynet combines the speed and cost-benefits of machine translation with the authenticity and precision that only a native speaker can deliver.

KantanMT is part of the Keywords Studios Group, the largest provider of global services to the video games and media and entertainment industries. It has offices in over 50 locations and delivers services to 23 of the top 25 games companies.

KantanMT is based in the INVENT Building, DCU Campus, Dublin 9, Ireland.