Bio

Marjorie Freedman's research spans natural language processing, machine learning, multimodal information understanding, and trustworthy AI. She and her group develop systems that transform complex, multilingual, and multimodal information into structured knowledge that can support human understanding and decision-making.

Her work has addressed information extraction, knowledge graph construction, complex event and temporal reasoning, human-machine collaboration, and the detection of contradictions and other anomalies in data. More recently, she has focused on the evaluation of chatbot responses, including the closely connected challenges of defining what constitutes a good response, developing automated methods to determine if those criteria have been met, and creating annotated datasets to validate those methods.

She has led large, interdisciplinary research efforts funded by DARPA, IARPA, and ARPA-H. She is particularly interested in combining rigorous evaluation with human-centered system design so that AI can augment expert judgment while remaining reliable, transparent, and responsive to human needs.