Progress of the PRINCIPLE Project: Promoting MT for Croatian, Icelandic, Irish and Norwegian

Progress of the PRINCIPLE Project: Promoting MT for Croatian, Icelandic, Irish and Norwegian Petra Bago,1 Jane Dunne,2 Federico Gaspari,2 Andre Kåsen,3 Gauti Kristmannsson,4 Helen McHugh,2 Jon Arild Olsen,3 Dana D. Sheridan,5 Páraic Sheridan,5 John Tinsley,5 Andy Way2 1 Faculty of Humanities and Social Sciences, University of Zagreb, 10000 Zagreb, Croatia 2 ADAPT Centre, School of Computing, Dublin City University, Dublin 9, Ireland 3 National Library of Norway, Henrik Ibsens gate 110, 0203 Oslo, Norway 4 University of Iceland, Saemundargata 2, 101 Reykjavik, Iceland 5 Iconic Translation Machines, Invent Building, Dublin City University, Dublin 9, Ireland pbago@ffzg.hr, {jane.dunne,federico.gaspari,helen.mchugh, andy.way}@adaptcentre.ie, {andre.kasen,jon.olsen}@nb.no, gautikri@hi.is, {dana,paraic,john}@iconictranslation.com Abstract This paper updates the progress made on the PRINCIPLE project, a 2-year action funded by the European Commission un- der the Connecting Europe Facility (CEF) programme. PRINCIPLE focuses on col- lecting high-quality language resources for Croatian, Icelandic, Irish and Norwe- gian, which have been identified as low- resource languages, especially for build- ing effective machine translation (MT) systems. We report initial achievements of the project and ongoing activities aimed at promoting the uptake of neural MT for the low-resource languages of the project. 1 Background PRINCIPLE is a 2-year initiative that started in September 2019, funded by the European Com- mission under the Connecting Europe Facility (CEF) programme. The project is coordinated by the ADAPT Centre at Dublin City University (DCU, Ireland), and the consortium includes the Faculty of Humanities and Social Sciences of the University of Zagreb (Croatia), the National Li- brary of Norway in Oslo, the University of Ice- land in Rejkyavik, and Iconic Translation Ma- chines (Ireland). PRINCIPLE focuses on the identification, collection and processing of high- quality language resources (LRs) for Croatian, © 2020 The authors. This article is licensed under a Creative Commons 3.0 licence, no derivative works, at- tribution, CC-BY-ND. Icelandic, Irish, and Norwegian (covering both varieties of Bokmål and Nynorsk), which are se- verely under-resourced. The uptake of machine translation (MT) for these languages has been hampered so far by the lack of extensive high- quality LRs that are required to build effective systems, especially parallel corpora. PRINCIPLE aims to improve LR collection efforts in the re- spective languages, prioritising the two strategic Digital Service Infrastructures (DSIs)1 of eJustice and eProcurement. The LRs assembled and cu- rated in PRINCIPLE will be validated to demon- strate improved MT quality, and will be uploaded via ELRC-SHARE to enhance MT systems pro- vided by eTranslation, that are available to public administrations in Europe, thus promoting lan- guage equality for low-resource languages. Way and Gaspari (2019) introduced the PRINCIPLE project at its start, giving a high- level overview of its main objectives, along with the planned activities and the overall approach to data collection and validation. They also ex- plained its position within the wider eco-system of related, recently finished CEF projects such as iADAATPA (Castilho et al., 2019) and ELRI.2 This paper provides an update on the progress of PRINCIPLE, focusing on its initial achievements and describing ongoing activities, especially in terms of engaging …

Content truncated to 500 words