Constrained Entity Selection under Partial Knowledge (CES-PK) for LLM-Based Knowledge Graph QA

Summary

Large language models are increasingly used for knowledge graph question answering (KGQA), but can fail to correctly ground answers in the underlying graph. Existing LLM-based KGQA approaches either rely on full semantic parsing into executable queries such as SPARQL, which is brittle in practice due to complex schemas or incompleteness of real-world knowledge graphs, or on LLM reasoning and answer generation over KGs, which is more robust but lacks formal guarantees. This paper by Emanuel Kitzelmann (arXiv:2608.24824) studies a complementary setting in which candidate answers are generated by an LLM-based system and then verified using lightweight symbolic constraints derived from the question. The authors introduce Constrained Entity Selection under Partial Knowledge (CES-PK), a problem formulation focused on eliminating invalid answers and providing symbolic support for valid ones without constructing executable logical forms. To account for incomplete knowledge graphs, CES-PK adopts a three-valued constraint semantics (satisfied, violated, unknown), avoiding false rejections under the open-world assumption.