The ISEE framework uses a scoring system to identify ambiguous database field descriptions. It gathers missing domain knowledge through interactive loops to improve downstream agent performance. This targeted enrichment helps LLM-based agents execute entity-linking and data retrieval tasks more accurately by bridging the gap between raw data and user context.