# Q-GNN: Query-Conditioned Graph Neural Networks with Type Awareness for Knowledge Graph Completion

Source: [Q-GNN: Query-Conditioned Graph Neural Networks with Type Awareness for Knowledge Graph Completion](<https://www.sota2.com/research/paper/6a22589886173ad9c1468b2b>)

Collection index: [llms.txt](<https://www.sota2.com/research/llms.txt>)

Data updated: 2026-06-06T05:00:04.316Z

Authors: Dongxiao He, Ruqiong Zhang, Zhizhi Yu, Ling Ding, Di Jin, Guangquan Xu, Zhiyong Feng

Published: 2026-06-04T03:07:22.000Z

## Abstract

Knowledge Graph Completion (KGC) aims at predicting missing triplets from incomplete knowledge graphs, which is crucial for downstream applications. Recently, Graph Neural Network (GNN)-based methods have achieved remarkable success by performing message passing over query-centered local subgraphs. However, in practice, a query is jointly defined by both the entity and the relation, with both carrying information indispensable for reasoning, yet these methods rely solely on the query relation as the guiding signal, while the information inherent in the query entity is not leveraged to guide inference - the entity serves merely as a structural anchor for subgraph extraction. To this end, we incorporate query entity information into the reasoning process from two perspectives: the first is structural context, i.e., the neighboring structure and relation patterns around the entity, which is encoded by a dedicated context encoder and used to modulate messages; the second is semantic type of the entity, inferred by a large language model, which is incorporated into attention computation and final scoring to provide type-level prior constraints. Together, these two sources of information enable the reasoning process to be guided by both the query relation and the query entity. Experimental results on standard benchmarks demonstrate the effectiveness of the proposed Q-GNN.

## Recorded benchmark results

A research result applies to its stated dataset, split, metric and evaluation configuration. Compare like-for-like settings, preserve the metric's unit and direction, and distinguish paper-reported results from independent reproduction. The best result recorded here is not a claim of exhaustive, current global coverage.

- [Knowledge Graph Completion on WN18RR standard (transductive)](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-wn18rr-standard-transductive.md>): Knowledge Graph Completion; WN18RR standard (transductive); MRR: 57.2.

- [Knowledge Graph Completion on FB15k-237 (standard transductive)](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-fb15k-237-standard-transductive.md>): Knowledge Graph Completion; FB15k-237 standard (transductive); MRR: 0.408.

- [Knowledge Graph Completion on NELL-995](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-nell-995.md>): Knowledge Graph Completion; NELL-995; MRR: 0.57.

- [Inductive Knowledge Graph Completion on WN18RR V1](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-wn18rr-v1.md>): Inductive Knowledge Graph Completion; WN18RR V1; MRR: 0.714.

- [Inductive Knowledge Graph Completion on WN18RR V2](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-wn18rr-v2.md>): Inductive Knowledge Graph Completion; WN18RR V2; MRR: 0.705.

- [Inductive Knowledge Graph Completion on WN18RR V3](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-wn18rr-v3.md>): Inductive Knowledge Graph Completion; WN18RR V3; MRR: 45.1.

- [Inductive Knowledge Graph Completion on WN18RR V4](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-wn18rr-v4.md>): Inductive Knowledge Graph Completion; WN18RR V4; MRR: 65.9.

- [Inductive Knowledge Graph Completion on FB15k-237 V1](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-fb15k-237-v1.md>): Inductive Knowledge Graph Completion; FB15k-237 V1; MRR: 46.

- [Inductive Knowledge Graph Completion on FB15k-237 V2](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-fb15k-237-v2.md>): Inductive Knowledge Graph Completion; FB15k-237 v2; MRR: 49.5.

- [Inductive Knowledge Graph Completion on FB15k-237 (V3)](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-fb15k-237-v3.md>): Inductive Knowledge Graph Completion; FB15k-237 v3; MRR: 46.7.

- [Inductive Knowledge Graph Completion on FB15k-237 V4](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-fb15k-237-v4.md>): Inductive Knowledge Graph Completion; FB15k-237 v4; MRR: 45.8.

- [Inductive Knowledge Graph Completion on NELL-995 V4](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-nell-995-v4.md>): Inductive Knowledge Graph Completion; NELL-995 v4; MRR: 33.6.

- [Knowledge Graph Completion on WN18RR V2](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-wn18rr-v2.md>): Knowledge Graph Completion; WN18RR V2; Hits@10: 80.8.

- [Knowledge Graph Completion on WN18RR V4](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-wn18rr-v4.md>): Knowledge Graph Completion; WN18RR V4; Hits@10: 73.8.

- [Knowledge Graph Completion on FB15k-237 V1](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-fb15k-237-v1.md>): Knowledge Graph Completion; FB15k-237 V1; Hit@10: 61.3.

- [Knowledge Graph Completion on FB15k-237 V2](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-fb15k-237-v2.md>): Knowledge Graph Completion; FB15k-237 v2; Hit@10: 69.3.

- [Knowledge Graph Completion on FB15k-237 V3](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-fb15k-237-v3.md>): Knowledge Graph Completion; FB15k-237 v3; Hit@10: 63.8.

- [Knowledge Graph Completion on FB15k-237 V4](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-fb15k-237-v4.md>): Knowledge Graph Completion; FB15k-237 v4; Hit@10: 65.5.

- [Knowledge Graph Completion on NELL-995 V1](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-nell-995-v1.md>): Knowledge Graph Completion; NELL-995 V1; Hits@10: 89.6.

- [Knowledge Graph Completion on Family](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-family.md>): Knowledge Graph Completion; Family; MRR: 99.1.

- [Knowledge Graph Completion on WN18RR V1](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-wn18rr-v1.md>): Knowledge Graph Completion; WN18RR V1; Hits@10: 81.1.

- [Knowledge Graph Completion on WN18RR V3](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-wn18rr-v3.md>): Knowledge Graph Completion; WN18RR V3; Hits@10: 55.

- [Inductive Knowledge Graph Completion on NELL-995 V2](<https://www.sota2.com/research/sota/inductive-knowledge-graph-completion-on-nell-995-v2.md>): Inductive Knowledge Graph Completion; NELL-995 V2; MRR: 37.4.

- [Knowledge Graph Completion on NELL-995 V3](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-nell-995-v3.md>): Knowledge Graph Completion; NELL-995 V3; Hit@10: 58.4.

- [Knowledge Graph Completion on NELL-995 V2](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-nell-995-v2.md>): Knowledge Graph Completion; NELL-995 V2; Hit@10: 53.2.

- [Knowledge Graph Completion on NELL-995 V4](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-nell-995-v4.md>): Knowledge Graph Completion; NELL-995 v4; Hit@10: 49.4.

- [Knowledge Graph Completion on UMLS](<https://www.sota2.com/research/sota/knowledge-graph-completion-on-umls.md>): Knowledge Graph Completion; UMLS; Hits@10: 0.99.

## Original sources and implementation

[Original paper](<https://arxiv.org/abs/2606.05639>)

[PDF](<https://arxiv.org/pdf/2606.05639.pdf>)
