Translation
Scientific content should travel across languages without losing terminology, nuance or traceability to the source paper.
CHIST-ERA Science in Your Own Language
Scientific knowledge, adapted to the language of its audience.
MUTASK develops AI-driven methods that transform scholarly papers into multilingual, audience-aware and multimodal explanations while preserving scientific fidelity.
What MUTASK Works On
Transposition
The same research result should be explainable to different audiences, from specialists and students to journalists, teachers and citizens.
Transformation
A paper can become summaries, scripts, visuals, narrated video segments, subtitles and interactive clarifications.
From One Paper To Many Useful Outputs
MUTASK treats a scientific paper as a structured knowledge object, not just as a block of text. The project lifecycle connects:
- paper intake, parsing and extraction of sections, figures, tables, equations, metadata and references;
- semantic indexing, knowledge graphs, content selection and translation;
- audience modelling, pedagogical adaptation and dissemination strategy;
- media production, interactive video and question-answer support;
- expert review, comprehension checks and release decisions.
First Demonstrator
The first demonstrator is intentionally small and testable: one scientific paper, one structured paper package, one audience profile, one adapted explanation path and one media output that can be reviewed for fidelity, clarity and usefulness.
This controlled path lets the consortium expose real technical and pedagogical problems before scaling to more papers, languages, audiences and media formats.
Consortium
MUTASK brings together expertise in document processing, multimedia systems, computational linguistics, machine translation, media education, audience research and strategic communication from Poland, France and Switzerland.