Elizabeth Clark

Elizabeth Clark is a researcher at the University of Washington, specializing in Natural Language Processing (NLP) and Computational Linguistics. Her work focuses on improving text generation systems through rigorous evaluation frameworks and datasets.

• Co-developer of the SEAHORSE dataset for multilingual summarization evaluation
• Principal investigator in context-aware rewriting and dialect-robust metrics
• Contributor to GEM benchmark standardization for NLG evaluation

Her research explores challenges in:

• Human evaluation reliability for machine-generated text
• Counterfactual reasoning in narratives
• Contextual rewriting to enhance stylistic coherence

Clark has published extensively in top conferences like ACL, EMNLP, and NAACL, often collaborating with Noah A. Smith and other NLP experts. Her work highlights critical gaps in automated evaluation metrics and proposes innovative solutions like CtxSimFit and NANO for robust assessment.