Pan du

Background

About Pan du

My research focuses on applying Machine Learning (ML) and Artificial Intelligence (AI) techniques to Information Retrieval (IR) and Natural Language Processing (NLP) tasks. This includes large language models (LLMs), question answering (QA), dense retrieval (DR), learning to rank (L2R), text summarization, and recommendation systems. I design and develop advanced retrieval and ranking systems, with a particular emphasis on legal language modeling, legal retrieval, document ranking, topic modeling, sentiment analysis, and social network analysis. My work bridges both theoretical research and real-world applications. Currently, I lead a team of scientists and engineers in developing legal language models (Legal LMs) and large-scale legal dense retrieval and ranking solutions. Our models have significantly outperformed strong baselines, achieving a 17% improvement in retrieval performance and a 15% enhancement in ranking. Their effectiveness has been validated across multiple projects and applications. Additionally, I spearheaded the design and deployment of a customized Retrieval-Augmented Generation (RAG) solution for legal professionals, which was released last month and has received outstanding customer feedback. Beyond product innovation, I actively contribute to the academic community by serving on program committees for leading IR and NLP conferences, including SIGIR, KDD, WSDM, CIKM, ECIR, ACL, EMNLP, NAACL, and COLING. I also file patents on system design methodologies and publish research in top-tier IR and NLP conferences and journals.