Natural Language Processing

We aim to conduct cutting-edge research and become a local hub in Asia in natural language processing (NLP) and language technology. Geographically, we are naturally drawn towards language problems and challenges in the region which might otherwise be overlooked in the research community. Our goal is to not only create new tools and knowledge agonistic to low-resource languages in the region, but also to practically create the very best NLP technology for Vietnamese. Consequently, we are pushing new state-of-the-arts in low-resource language problems, language modeling and translation, conversational AI, information extraction, and the like.

New technology requires new fundamental research. To this end, our team collaborates with the Machine Learning team to work on foundations of machine learning for NLP such as self-supervised learning, adversarial learning, multi-task learning, graph neural networks and knowledge graph, and also collaborates with the Computer Vision team for multimodal research in vision and language.

The NLP team has helped boost the global visibility of VinAI by establishing a strong collaborator network with prominent researchers all over the world, for example, from the University of Oregon in the USA, Nanyang Technological University in Singapore, the University of Melbourne and Monash University in Australia. We achieved substantial research outputs with papers published at top-tier NLP/AI conferences, under a wide range of, but not limited to, the following topics:
- Foundation models and Large language models
- Text classification and summarization
- Text and speech translation
- Question answering and dialogue systems
- Spoken language understanding
- Tagging, syntactic and semantic parsing
- Relation and event extraction
- Knowledge graph embedding
- Language grounding to vision
- Resources and evaluation