Our research encompasses all aspects of NLP, from modeling basic linguistic phenomena to designing practical text processing systems, and developing new machine learning methods.
A central theme of our research is developing creative new algorithms for processing text and other information (images, software, chemical reaction) to solve important social problems and expand the capabilities of the field. We are particularly interested in pushing the NLP boundaries to new tasks. We were the first to demonstrate statistical decipherment of a dead language, develop an automatic Civilization player that learns from text manuals, and design a system for generating Wikipedia articles. Current research includes designing NLP systems capable of performing competitively without requiring expensive and slow to obtain manual annotations. Instead, we are exploring alternative learning approaches which can benefit from other sources of supervision. Examples include grounding language in control applications, cross-lingual transfer, and learning with declarative rules.
Interested in working with the Natural Language Processing Group?
Researchers
If you would like to contact us about our work, please scroll down to the people section and click on one of the group leads' people pages, where you can reach out to them directly.
UROP Opportunities
If you would like to contact us about our work, please scroll down to the people section and click on one of the group leads' people pages, where you can reach out to them directly.
We propose a novel aspect-augmented adversarial network for cross-aspect and cross-domain adaptation tasks. The effectiveness of our approach suggests the potential application of adversarial networks to a broader range of NLP tasks for improved representation learning, such as machine translation and language generation.
“Co-LLM” algorithm helps a general-purpose AI model collaborate with an expert large language model by combining the best parts of both answers, leading to more factual responses.
New CSAIL research highlights how LLMs excel in familiar scenarios but struggle in novel ones, questioning their true reasoning abilities versus reliance on memorization.
Three neurosymbolic methods help language models find better abstractions within natural language, then use those representations to execute complex tasks.
A team of scientists demonstrated that AI-risk models, paired with AI-designed screening policies, can offer significant and equitable improvements to cancer screening
This week it was announced that MIT professor and CSAIL principal investigator Regina Barzilay was named an Association for the Advancement of Artificial Intelligence (AAAI) Fellow for her research contributions to the field of artificial intelligence (AI).
