data-ingestion

This paper presents a system that automatically discovers an executable schema from raw multi-source data and uses it for knowledge graph construction and query-time retrieval, improving over baselines on QA benchmarks.