LLM and OCR-Based Document Understanding

This work included evaluating language-model platforms, analyzing OCR robustness, extracting structured JSON, visually associating extracted results with document regions, and supporting data normalization and validation with reference data.