|
Newsletter #17 – July 2022
|
|
|
|
Dear reader,
Last month saw not only this year's META-FORUM (see our previous newsletter for some highlights), but also the conclusion of the ELE project. ELE 2, which is already underway as of last Friday, will take ELE’s work further, pursuing the shared goal of European digital language equality by 2030. Stay tuned for the open call for contribution projects, to be published in September.
On the ELG front, the ELG consortium met virtually last week for the final review meeting and we couldn’t be happier with the highly positive feedback. This bodes well for ELG’s future as a legal entity and as the go-to place for European LT services and resources.
As this newsletter goes out, New Trends in Translation and Technology (NeTTT) is underway in Rhodes, with colleagues from our partner, DCU, amongst the chairs and organisers. We wish all participants a successful and enjoyable rest of the conference!
On a final note, this newsletter will be returning to a monthly publication schedule. We look forward to continuing to provide you with updates from the European technology community.
With best regards
Georg Rehm
|
|
|
Language Technology and NLP in the news
|
|
|
- “Can Machine Learning Translate Ancient Egyptian Texts?” – Sapiens, 9 June 2022
- “Why AI fairness tools might actually cause more problems” – Protocol, 18 June 2022
- “If AI chatbots are sentient, they can be squirrels, too” – The Register, 20 June 2022
- “Offload Shiny’s Workload: COVID-19 processing for the WHO/Europe” – R-Bloggers, 21 June 2022
- “SAP AI lead: ‘Europe has fallen behind in AI commercialisation’” – Silicon Republic, 22 June 2022
- “What’s new in Microsoft Azure’s NLP AI services” – InfoWorld, 22 June 2022
- “Cerebras Systems sets record for largest AI models ever trained on one device” – VentureBeat, 22 June 2022
- “Natural Language Processing and “Mindful” AI Drive More Sophisticated Bad Bot Attacks” – Security Boulevard, 22 June 2022
- “Open-source language AI challenges big tech’s models” – Nature, 22 June 2022
- “Zoom Launches Translation Feature — But Only for Business Users” – Slator, 23 June 2022
- “Multimodal Learning: A New Frontier in Artificial Intelligence” – Techopedia, 23 June 2022
- “A Phenotype Ontology for Autism Spectrum Disorder Was Created By Using Natural Language Processing on Electronic Health Records” – Physician’s Weekly, 24 June 2022
- “Google’s powerful AI spotlights a human cognitive glitch: Mistaking fluent speech for fluent thought” – The Conversation, 24 June 2022
|
|
|
The LaMDA chatbot might say yes when asked if it is a squirrel.
|
|
- With the summer heat most of Europe is currently experiencing, we all feel like cutting corners. But can machine translation models also feel an urge for creative laziness?
- Last week’s final project review meeting for the European Language Grid was a huge success (see below) – so much so, that we don’t even mind looking at another Zoom screenshot.
- Hangouts, Chat, Talk, Rooms, Spaces - Google’s branding for interactive platforms does get confusing. Wasn’t there also a plus, or are we thinking of the Swiss flag (see below again)?
- Apart from ELG becoming a legal entity (correct, a third reason to see below), guess who else found their way into a non-profit organisation?
|
|
3.5 years have passed since the start of the European Language Grid project, during a time when the world still felt very different. Considering the many challenges that arose during the last 42 months, we are even happier about the feedback that ELG received during its final review meeting: Last Tuesday, 28 June, the consortium met with Project Officer Philippe Gelin (DG CONNECT, European Commission) and two external reviewers, who were presented with all the achievements of this, as one reviewer put it, “amazing project”. Almost all of the KPIs, defined in early 2018 (!), were met or exceeded by a large margin and the latest release of the platform works exactly as envisioned all those months ago.
Now, after the end of the official project runtime, the ELG platform will be hosted, managed and maintained through a legal entity – a registered non-profit association under German law to be established in the second half of 2022. The organisation will gradually explore new models and areas of business. In short: ELG stays alive, so keep using it as your platform and marketplace for LT services and resources and make sure to spread the word within the European LT community! In this newsletter, we will report about all updates and new developments concerning the platform.
Apart from our own newsletter, we were also invited to write a recap of our successful META-FORUM conference for the European Language Resource Coordination (ELRC), which was published on their website last Thursday. For a short look back at the first META-FORUM on-site in three years, check out the ELRC news section.
|
|
Selected new tools and resources on the
European Language Grid
|
|
|
|
ZRCola – ZRCola is an input system designed mainly, although not exclusively, for linguistic use. It allows the user to combine basic letters with any diacritic marks and insert the resulting complex characters into the texts with ease. The system is comprised of an input program and a font, which can also be installed separately. The font is based on the Unicode standard and includes a vastly enlarged set of Latin, Cyrillic and other characters for Slavic writing systems in the Private Use Area. The tool was harvested from CLARIN.SI and provided by the Fran Ramovš Institute of the Slovenian Language on 27 June, 2022.
|
|
Limecraft – Limecraft offers online workspaces for professionals in creative industries, including Media Asset Management (MAM) and Workflow Automation solutions. News agencies, producers, and broadcasters worldwide rely on Limecraft to manage their material, thereby automating manual work so they can concentrate on better storytelling.
Using audio transcription, Limecraft automates tedious tasks like video logging and storyboarding. Extended with Natural Language Processing, Limecraft also automates the subtitling and localisation processes.
Using Limecraft, journalists, media managers and editors can produce significantly more in less time. It allows them to create and dispatch multiple versions at the same time, thereby maximising their target audience.
|
|
The end of June saw the official end of ELE, but also the launch of its successor project, ELE 2. ELE 2 will build on the foundations laid by ELE and consists of seven partners: Dublin City University (the coordinator of the project), the German Research Center for Artificial Intelligence (DFKI, co-coordinator of the project once again), Charles University, University of the Basque Country (EHU), Athena Research & Innovation Center, the European Federation of National Institutions of Language (EFNIL), and the European Language Equality Network (ELEN).
These partners will be working together to further develop the strategic research and implementation agenda (SRIA) developed by ELE for achieving digital language equality in Europe by 2030. This includes creating a list of the most urgently needed language resources, technologies and services to ensure adequate technological support for all of ELE 2’s languages.
Another highlight of ELE 2 are the SRIA contribution projects, which will produce concrete use cases and best-practice examples of language resource development and language technology implementations in relevant industry sectors and areas of life. The open call for contribution projects will be published in September, so keep an eye out!
The results of the projects as well as the updated SRIA, will be presented to the language technology community for discussion and feedback. This will take place at the ELE 2 conference, which will happen as part of META-FORUM 2023, just before the project concludes in June 2023.
FYI: New project, same details. Information about the work of ELE will continue to be available on the website, in this newsletter, and on our Twitter and LinkedIn.
|
|
Much like its flag, Switzerland’s multilingual setup is a big plus for the country. But German, French and Italian are not the only languages spoken in the central-European country: While the three dominant languages rank pretty well on our Digital Language Equality Dashboard, the outlook is quite different for Romansh, an official Swiss language spoken in the canton of Grisons. Now, for the first time, machine translation is available to and from Romansh, thanks to the AI translation provider TextShuttle and the Rhaeto-Romanic media company Radiotelevisiun Svizra Rumantscha (RTR). You can find all details about this important development in this Slator article or try out the translation service yourself. Bun divertiment!
|
|
The next ELT newsletter will be sent out on 2 August 2022. Until then, follow our ELT social media accounts (as linked below) for the latest news!
|
|
|
|
|
