Brick-DICL: Dynamic In-Context Learning for Automated Brick Schema Classification

In three linesBrick-DICL introduces a two-stage dynamic in-context learning framework for automated Brick schema classification of BMS points (936 classes). Combines metadata-RAG and class-RAG to enhance LLM domain knowledge, with multi-LLM filtering to reduce manual verification effort.

Summary generated by Claude — human-verified