zelph

When automated reasoners draw conclusions over open knowledge graphs such as Wikidata, the inference is typically a black box: a verdict comes out, but the proof stays hidden inside the tool. zelph is an open-source semantic network engine built on the opposite principle: facts, rules and derivations share one and the same graph representation, so every inference step is inspectable, machine-readable data. zelph already performs logical inference over the complete Wikidata knowledge graph on a single machine and has surfaced thousands of consistency violations in collaboration with the Wikidata Ontology Cleaning Task Force. This project turns that capability into domain-agnostic infrastructure for the Linked Open Data commons: streaming import adapters for RDF/N-Triples and JSON-LD, an open proof-export format that publishes complete derivation chains as Linked Data, a reusable library of constraint rules covering common ontology semantics, and a web-based proof explorer that lets anyone trace a derived fact back to its evidence. Any community maintaining an open knowledge graph will be able to audit its logical consistency — with full transparency over every reasoning step.