Lost in meaning – found in translation: Natural language understanding with multilingual data

Jörg Tiedemann

University of Helsinki, Department of Digital Humanities

Natural language understanding is the “holy grail” of Computational linguistics and a long-term goal in research on artificial intelligence. The aim of the FoTran project is to develop models that learn to understand human languages by training on implicit information given by large collections of human translations. Translations are considered as alternative explanations providing additional views on information encoded in natural language. In the project we apply massively parallel data sets to acquire language-agnostic meaning representations that can be used for reasoning with natural languages and for other downstream tasks that require a deep understanding of the linguistic input. In my presentation, I will discuss our recent studies with multilingual machine translation models and cross-lingual sentence representations.


More information about the FoTran project:
https://www.helsinki.fi/en/researchgroups/natural-language-understanding-with-cross-lingual-grounding