Charles L. A. Clarke

Information retrieval researcher working on evaluation methodology, including the reliability of LLM-generated relevance judgments as a substitute for human assessment.

Known For

• Benchmarking LLM-based Relevance Judgment Methods (arXiv:2504.12558, 2025) with Negar Arabzadeh — establishes that no single LLM judgment paradigm is best on both label agreement and system-ranking agreement

• Co-organizer of the LLMJudge challenge at the LLM4Eval workshop, SIGIR 2024

Articles

• Benchmarking LLM-based Relevance Judgment Methods

Related Datasets

• LLMJudge

Related Concepts

• LLM as Judge
• Search Evaluation
• Pointwise Relevance Evaluation

People

• Negar Arabzadeh — co-author