HyGRAIL: Cost-Aware and Evidence-Grounded Scientific Hypothesis Discovery over Knowledge Graphs

HyGRAIL: экономичная и основанная на доказательствах генерация научных гипотез на графах знаний
Yihang Sun, Zhihan Zhu, Zhiyuan Jiang, Jingyi Ge, Zixuan Li, Jiaxuan You
2026-09-02

evidence-grounded verificationheterogeneous graph neural networksknowledge graphslarge language model reviewscientific hypothesis discovery
Scientific knowledge graphs organize entities and relations extracted from scientific literature, but they remain inherently incomplete. Missing typed links in such graphs can therefore represent plausible scientific hypotheses, such as unexplored associations between materials and applications. However, scientific hypothesis discovery is challenging because true discoveries are extremely sparse among typed candidate pairs: graph neural networks (GNNs) are efficient but unreliable for ambiguous cases, while large language models (LLMs) are knowledgeable but too costly to apply exhaustively and are not naturally grounded in graph structures. We propose HyGRAIL, a cost-aware and evidence-grounded framework that combines heterogeneous GNN triage with LLM-based hypothesis review. HyGRAIL first uses a GNN to score candidate hypotheses and identify a validation-calibrated ambiguous region, routing only graph-uncertain cases to LLM review. For each routed hypothesis, HyGRAIL retrieves node-level associations and multi-hop relational paths from the knowledge graph (KG), then converts this structured evidence into natural language through template-based or LLM-based naturalization. An LLM review agent finally judges each hard hypothesis using the naturalized evidence and validation-selected decision criteria. On MatKG, HyGRAIL achieves the best F1 score of 0.429, improving over the strongest prior baseline by 0.242 F1 points and over the GNN-only baseline by 0.322. Meanwhile, GNN triage reduces the LLM call rate by 54.36% on average. Ablation studies further show that retrieved graph evidence is crucial for reliable hypothesis verification and that compact, two-sided evidence is more effective than simply increasing retrieval quantity.
1
Ablations show that retrieved graph evidence is essential for reliable verification, while compact two-sided evidence outperforms simply increasing retrieval quantity.
2
HyGRAIL combines heterogeneous GNN triage with evidence-grounded LLM review to discover missing typed links as scientific hypotheses in incomplete knowledge graphs.
3
HyGRAIL grounds LLM decisions in retrieved node associations and multi-hop knowledge-graph paths converted into natural-language evidence.
4
On MatKG, HyGRAIL achieves the best F1 score of 0.429, improving by 0.242 points over the strongest prior baseline and 0.322 over GNN-only classification.
5
The framework routes only validation-calibrated, graph-ambiguous candidates to LLM review, reducing the average LLM call rate by 54.36%.

missing typed links representing candidate scientific hypotheses in scientific knowledge graphs, particularly material–application associations

cost-aware, evidence-grounded hypothesis verification and discovery, including graph-based triage, LLM review, and the effectiveness of retrieved evidence

Publication Details
Publication Date
2026-09-02
Journal
Publisher
ISSN
Access Type
Author Information
Authors
Yihang Sun
Zhihan Zhu
Zhiyuan Jiang
Jingyi Ge
Zixuan Li
Jiaxuan You
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat →
Make a presentation
100%