ExpertQA is a benchmark dataset and evaluation framework for assessing the factual accuracy and source attribution of long-form question-answering systems across specialized academic and professional domains. Developed to measure how effectively language models handle complex, high-stakes inquiries in fields such as medicine, law, and engineering, it consists of questions authored by subject-matter experts across dozens of disciplines alongside detailed answers evaluated and refined for verifiable claim attribution. By prioritizing expert judgment over crowdsourced annotations, it provides a standard for testing whether automated systems can generate domain-specific information that is factually reliable and grounded in credible sources.
