LLM-based schema mapping can be made more scalable and robust through sampled prompts, bidirectional confidence aggregation, and rule chunking, letting a smaller open-source model match GPT-4-based performance on MIMIC and Synthea benchmarks.
Title resolution pending
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.DB 1years
2025 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Towards Scalable Schema Mapping using Large Language Models
LLM-based schema mapping can be made more scalable and robust through sampled prompts, bidirectional confidence aggregation, and rule chunking, letting a smaller open-source model match GPT-4-based performance on MIMIC and Synthea benchmarks.