Coronavirus (Covid-19) Information

Bangor University Machine Translation KTP graded as Outstanding by Innovate UK

KTP Associate Myfyr Prys at work at Cymen.KTP Associate Myfyr Prys at work at Cymen.Bangor University and Cymen Cyf have been awarded an A (Outstanding) rating for their recently completed Knowledge Transfer Partnership (KTP). Cymen is a translation company located in Caernarfon, and is one of the largest private sector employers of graduates in Gwynedd. The KTP focused on developing domain-specific machine translation between English and Welsh, using Cymen’s large archive of translated documents as training data.

Dewi Bryn Jones, of the University’s Language Technologies Unit (LTU) at Canolfan Bedwyr, who was Academic Supervisor on the project, explained:

“Our research aims at making language technologies accessible and freely available for supporting the Welsh language and Welsh economic activity. This KTP was an exciting opportunity for us to transfer and exploit our expertise in domain specific machine translation with Cymen’s vast archive of legacy translations. Working closely with the KTP associate, Myfyr Prys, we were able to demonstrate and present at a recent European machine translation academic conference in Dublin that Cymen’s bespoke engines gave much better results than larger general purpose translation engines such as those found on the web.”

Manon Cadwaladr, one of Cymen’s directors, said:

“We were pleasantly surprised by the quality of the output which we have now incorporated into our translation pipeline, using it in conjunction with human translators. Contrary to perceptions that machine translation makes translators redundant, we have been able to expand our workforce, employing an extra five translators as a result. We have increased our ability to produce high quality translations to tight deadlines, and both turnover and profitability have increased.”

Cymen and Bangor’s LTU have now been awarded a SMART Partnership grant by the Welsh Government to continue their collaboration, with a new emphasis on exploiting more recent developments in artificial intelligence and deep learning approaches for machine translation, as well as enabling home working by translators in remote, rural locations.

Congratulating Cymen and the LTU on the success of their project, Professor Iwan Davies, Vice-Chancellor of Bangor University said:

“‘A’ grades are very rarely given for these KTP projects and it is an exceptional achievement for both partners to have been involved in a project rated as being of outstanding quality. We are especially pleased that this collaboration has helped to develop Welsh language technology research and has been of benefit to the local economy. Given that the partners have won further funding through the SMART Partnership scheme, we now look forward to further successes in developing machine translation between English and Welsh.”

Minister for International Relations and Welsh Language, Eluned Morgan, said:

“Improved automated translation technology is a key objective of our Welsh Technology action plan. It is good to see innovation in the field at Bangor University and Cymen being recognised in this area.”

This partnership received financial support from the Knowledge Transfer Partnerships (KTP) programme . KTP aims to help businesses to improve their competitiveness and productivity through the better use of knowledge, technology and skills that reside within the UK knowledge base. This successful Knowledge Transfer Partnership project, funded by the Welsh Government and UK Research and Innovation through Innovate UK, is part of the government’s Industrial Strategy. 

The SMART Partnership project is part-funded by the Welsh Government. They offer financial support to innovative collaborative projects that require a range of expertise to help businesses grow, improve productivity and increase competitiveness.  The aim of SMART Partnerships is to increase the capability and capacity of Welsh businesses to undertake RD&I activity through knowledge transfer.

Publication date: 8 November 2019