معرفی
Hongrae Lee is a researcher specializing in database systems, natural language processing, and data mining. His work bridges structured data management with language models, focusing on tasks like natural language to SQL translation, similarity joins, and efficient data processing.
His research interests include:
- Database query optimization and similarity search
- Language model applications for text generation and hallucination correction
- Web data curation and structured data ecosystems
- Cloud storage optimization and distributed systems
Hongrae Lee's recent publications (2022-2023) highlight trends in large language models (LLMs) for dialogue applications, attributed text generation, and acronym disambiguation with weak supervision. Earlier work (2016-2007) established foundational techniques in database scalability, LSH-based similarity estimation, and geographical data thinning.
He has collaborated extensively with researchers at institutions like Google, Seoul National University, and University of British Columbia on projects such as WebTables, LaMDA, and Google Fusion Tables. His contributions span both theoretical advancements (e.g., variance-aware query optimization) and practical systems (e.g., CloudRAMSort, T5-based disambiguation).
