When Correct Edges Cannot Be Verified: A Provenance Gap in Incomplete KGQA and a Provenance-Favoring Completion Policy

2026-06-14

AI Summary

Existing knowledge graph question answering methods assume that the correctness of completed edges can be judged by textual verifiability, yet this assumption remains unvalidated. This work presents the first empirical finding that most correctly completed edges lack explicit textual support, and that textual verifiability often reflects provenance rather than factual accuracy. To address this, we propose TGComplete, a novel paradigm that balances precision and auditability through reasoning-breakpoint-guided evidence retrieval, a lightweight verification loop, and a provenance credibility–driven mechanism for edge adoption or abstention. Compared to the GoG baseline, TGComplete maintains answer accuracy while improving strict faithfulness of adopted edges by 3.1–7.4× and achieving edge precision of 15–21% (versus 3–14%), at the cost of a modest reduction in recall.

Abstract

Incomplete Knowledge Graph Question Answering (IKGQA) requires completing missing edges to continue reasoning. A growing line of work verifies completed edges against retrieved text, treating textual support as a proxy for edge quality. We ask a question that, to our knowledge, has not been systematically tested: does textual verifiability actually track correctness? Exploiting the gold deleted triples provided by the standard random-deletion protocol, we measure both. The finding is counterintuitive: among gold-correct completed edges, 76-96% have no supporting passage even under exhaustive retrieval, robustly across deletion rates (20%/40%), datasets (CWQ/WebQSP), and relation types (structural, commonsense, long-tail). Most Freebase-style facts simply do not occur as head-tail co-mentions in text. Textual faithfulness therefore measures provenance, not correctness -- separated by a paradigm-level gap no in-corpus retrieval closes. This reframes edge completion. Since most completed edges -- correct or not -- are causally redundant for the answer (95-97% of correct answers do not depend on any unsupported edge), the central question shifts from "is the edge correct?" to "admit or abstain under provenance uncertainty?" Within this framing we present TGComplete, a provenance-favoring admission policy that retrieves evidence at a reasoning breakpoint, verifies a candidate through a lightweight loop, and abstains when support is absent. Against the generate-to-complete baseline GoG, it attains higher edge precision against gold (15-21% vs 3-14%), with no statistically detectable EM loss and 3.1-7.4 times higher strict faithfulness of admitted edges -- at the cost of lower recall. We position TGComplete not as uniformly better, but as a principled point on a precision/provenance-recall trade-off, appropriate when auditability matters.

Problem

• Knowledge Graph Question Answering
• Incomplete Knowledge Graph
• Provenance Gap
• Textual Verifiability
• Edge Completion

Innovation

• provenance gap
• incomplete KGQA
• textual faithfulness
• edge completion
• abstention policy