Harvard NLP studies machine learning methods for processing and generating human language. We are interested in mathematical models of sequence generation, challenges of artificial intelligence grounded in human language, and the exploration of linguistic structure with statistical tools.

Our group's research publications and open-source projects have focused on text summarization, neural machine translation, visualizing recurrent neural networks, algorithms for shrinking neural networks, models for entity tracking in documents, multi-modal text generation, grammatical error correction, and new approaches for text generation.