CHIST-ERA intends to launch end of 2023 or beginning of 2024 a multilateral call on Machine Translation focusing on scientific documents: Science in your own language.
E.g., would an automatically translated scientific article be accepted at a top tier conference? Because the reasoning of an article can be dramatically altered by a single translation error, the topic encompasses several scientific challenges: Translation of complex documents (graphs, images…) with intricate cross-references, terminology and phraseology issues, document level translation, scientific domain adaptation etc.
Besides, the strategic objectives of the call are:
- Promote multilingualism in science;
- Support open evaluation methodology and experiment reproducibility.
Research funding organisations from Europe and beyond are welcome to join CHIST-ERA and participate in the call (contact: chistera@anr.fr).
Multilingualism in Science
With the topic Science in your own language, CHIST-ERA joins the agenda of UNESCO, the Coalition for Advancing Research Assessment (CoARA) and Helsinki Initiative to “promote multilingualism in scholarly communication”.
Advances in translation technologies offers an unprecedented opportunity to grasp the potential of multiple languages in Science:
- Open science to the citizens and the society;
- Take on board all research works whatever their language;
- Value research with regionally or nationally targeted impact;
- Make creativity and reasoning in research benefit from all language subtleties.
Researchers and citizens from English speaking countries also value the rise of such technologies, because science embraces an important share in non-English languages. By contributing to making access to science universal, multilingualism is aligned with the objectives of the ongoing Open Science transformation of the scientific landscape.
With more than 30 national or regional research funders on board, CHIST-ERA is well positioned to tackle this ambition.
Open Evaluation Methodology & Experiment Reproducibility
The systems and technologies developed by the funded projects to approach the objective of the call should be evaluated and ultimately compared.
Objective evaluation is not a triviality in Machine Translation, as in any other knowledge processing domains, and requires an in-depth expertise. While the research community has already, and with great success, endorsed this framework, its public support is still nascent in Europe, in spite of its demonstrated efficiency in terms of research management (see attached flyer).
For the past decade, CHIST-ERA has supported technology evaluation by promoting open evaluation methodology and experiment reproducibility in its yearly call. With this call, it aims at further deploying this ambition:
Standard Call + Technology Evaluation Instrument
A dialogue with the research community will be established to identify, in addition to the scientific objectives of the call, the metrics and protocols that will form the evaluation instrument.