Neural Extraction Framework: Enhancing Knowledge Graph Extraction via Neuro-Symbolic Integration

This project scales the DBpedia Neural Extraction Framework into a production-ready system by integrating a Neuro-Symbolic validation layer. Key innovations include an Axiomatic Consistency Checker to eliminate LLM hallucinations using DBpedia OWL constraints, and a high-throughput Redis-backed architecture capable of indexing the full 50GB Databus dump with sub-millisecond latency. By bridging probabilistic neural extraction with deterministic symbolic reasoning, the framework ensures the generation of high-precision, ontologically valid RDF triples for the DBpedia 2026 release.