ChemOntoRule Builds a Verifiable Symbolic Core for AI Chemistry Solving

Summary

Researchers introduce ChemOntoRule, a symbolic core for school-level chemistry problem solving. It combines a task-centric ontology, RDF/JSON representations, deterministic Python rules, and expert-coded fallbacks. The system matched 296 of 300 manually validated problems, but the authors caution that the evaluation set also shaped ontology construction, so the result measures coverage and internal consistency rather than independent generalization. A future design would use a language model mainly to translate natural-language questions into normalized task frames.