Research Statement — Dani Yogatama

I design algorithms for intelligent processing of natural language texts—for example, to extract factual information into a structured database (e.g., extracting headquarters locations, CEOs, and phone numbers of companies from text into a database) or to predict real-world events from text (e.g., scientific trends, disease outbreaks). These applications require models of text that scale to large datasets. I advance machine learning (ML) methods for natural language processing (NLP), focusing on large-scale sparse models that leverage expert-informed domain knowledge. In my research, I seek to answer the following questions: