Thien Huu Nguyen

My research explores mechanisms to understand human languages for computers so that computers can perform cognitive language-related tasks for us. Among others, I am especially interested in distilling structured information and mining useful knowledge from massive and multilingual human-written text of various domains.

Toward this end, our lab employs and designs effective learning algorithms for information extraction and text mining in natural language processing and data mining. We are currently focusing on deep learning algorithms to solve such problems. We are among the first groups that develop deep learning models and demonstrate their effectiveness for information extraction.

We are also targeting other language-related problems with deep learning, including reading comprehension, machine translation, natural language generation, chatbots and language grounding.

• FourIE: For a better idea about our research on information extraction, check out a demo for our recent neural information extraction system (performing joint entity mention detection, relation extraction, event detection, and argument role prediction) here.


• Trankit: a light-weight transformer-based toolkit for multilingual NLP that can process raw text and support fundamental NLP tasks for 56 languages. Trankit is based on recent advances on multilingual pre-trained language models, providing state-of-the-art performance for Sentence Segmentation, Tokenization, Multi-word Token Expansion, POS Tagging, Morphological Feature Tagging, Dependency Parsing, and Named Entity Recognition over 90 Universal Dependencies treebanks. Trankit can be installed and used easily with Python. Check out Trankit's documentation page for installation and usage. We also provide a demo and release the code for Trankit at our github repo.