Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction

Filip Kronström; Ross D. King

cs.LGAug 24, 2026

Hierarchy-Aware Semantic Losses for Knowledge Graph Link Prediction

Authors: Filip Kronström, Ross D. King

Organizations: Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Sweden · Department of Chemical Engineering and Biotechnology, University of Cambridge, United Kingdom

Abstract

Knowledge graphs are often accompanied by ontological class hierarchies that encode valuable semantic information, yet many link prediction methods either ignore such hierarchies or incorporate them indirectly through additional graph edges. Recent work introduced hierarchy-aware graph neural networks (GNNs), which use semantic losses derived from box embeddings to encourage satisfaction of subclass relationships during GNN-based representation learning. While this approach has shown promise for biological regression tasks, its effectiveness for knowledge graph link prediction has not been investigated. In this paper we evaluate hierarchy-aware semantic losses on link prediction across three benchmark datasets: AIFB, CoDEx, and BioKG. We combine graph neural network encoders with box-embedding-based semantic losses that encourage learned representations to better satisfy ontology-derived class hierarchies, and compare this approach to both standard link prediction models and models incorporating subclass relations as graph edges. Across all datasets, hierarchy-aware semantic losses significantly improve mean reciprocal rank (MRR) and consistently outperform models that incorporate hierarchy information through additional subclass edges. Relative to the baseline GNN models, MRR improved by 7.6%, 2.4%, and 15.5% on AIFB, CoDEx, and BioKG, respectively. Furthermore, semantic losses consistently outperform the alternative of augmenting the graph with subclass edges. These results are consistent with ontology-derived class hierarchies providing complementary information to graph structure, and suggest that encouraging hierarchical consistency through semantic losses is an effective and comparatively parameter-efficient mechanism for improving knowledge graph link prediction.

Figures & tables

Figure 1: Overview of the hierarchy-aware GNN architecture. Semantic losses are applied to the GNN encoder to produce embeddings encouraged to better satisfy subsumptions defined in ontologies. This is optimised jointly with a task-specific decoder, e.g., for link prediction.

\smaller Dataset	$\|\mathcal{E}\|$	$\|\mathcal{R}\|$	\smaller Train triples	\smaller Val/Test triples	\smaller Hierarchy entities	\smaller Subclass axioms	\smaller Coverage (%)	\smaller Max depth	\smaller Median depth
\smaller AIFB	2,431	36	30,102	416	172	3,890	100.0	5	2
\smaller CoDEx	77,951	69	551,193	30,622	1,986	69,974	80.2	22	10
\smaller BioKG	93,773	51	4,762,678	162,870	64,985	455,560	84.7	58	6

Table 1: Overview of dataset and hierarchy characteristics.

|\mathcal{E}|

and

|\mathcal{R}|

denote the number of entities and relations in the original KG. Hierarchy entities and subclass axioms denote the number of additional entities and subclass relations introduced by the hierarchy. Coverage indicates the percentage of KG entities represented in the hierarchy. Domain-specific hierarchy characteristics for BioKG are reported in tab:biokg.

Domain	$\|\mathcal{E}\|$	Hierarchy entities	Subclass axioms	Coverage (%)	Max depth	Median leaf node depth
function	45,085	618	93,040	100.0	16	6
disease	10,687	25,804	163,606	83.4	58	24
sideeffect	9,969	28,217	166,898	91.2	57	16
protein	17,499	2,040	21,906	82.9	6	2
drug	10,533	8,306	10,110	17.1	11	6

Table 2: Hierarchy characteristics for the five domains in BioKG.

		ComplEx		R-GCN / GraphSAGE
Dataset	Metric	Baseline	SC-Edges	Baseline	SC-Edges	SemLoss
AIFB	MRR	0.4056 \smaller $\pm$ 0.0085	0.4336 \smaller $\pm$ 0.0062	0.4758 \smaller $\pm$ 0.0237	0.4872 \smaller $\pm$ 0.0257	0.5118 \smaller $\pm$ 0.0151
AIFB	#Params (M)	1.32	1.35	0.51	0.51	0.51
CoDEx	MRR	0.3734 \smaller $\pm$ 0.0010	0.3700 \smaller $\pm$ 0.0014	0.3999 \smaller $\pm$ 0.0069	0.4035 \smaller $\pm$ 0.0074	0.4095 \smaller $\pm$ 0.0025
CoDEx	#Params (M)	39.95	40.96	6.63	6.74	6.69
BioKG	MRR	0.7440 \smaller $\pm$ 0.0044	0.6963 \smaller $\pm$ 0.0223	0.7232 \smaller $\pm$ 0.0785	0.7644 \smaller $\pm$ 0.0393	0.8354 \smaller $\pm$ 0.0144

Table 3: Link prediction performance (MRR) and number of trainable parameters. Results are reported as mean

\pm

standard deviation over 10 runs. All models trained with the semantic loss significantly outperform the corresponding baseline models (Welch’s two-sided t-test,

p<0.05

Figure 2: Evolution of domain-specific semantic losses during training on BioKG. Curves show the median semantic loss over 10 training runs and shaded regions indicate the interquartile range. Losses are reported separately for each ontology domain and GNN layer. The semantic losses decrease throughout training for all domains, indicating increasing satisfaction of ontology-derived hierarchy constraints, although the magnitude of improvement varies between domains. Layer 0 corresponds to the initial node embeddings prior to message passing.

Figure 3: Exploratory comparison between performance improvements obtained by hierarchy-aware semantic losses and hierarchy characteristics across the three evaluated datasets. The x-axis shows the improvement in mean reciprocal rank (

\Delta

MRR) relative to the corresponding baseline model. Hierarchy density is defined as the number of subclass axioms divided by the number of hierarchy entities. Hierarchy availability is defined as the number of hierarchy entities divided by the number of hierarchy-covered graph entities and reflects the amount of hierarchy information available per covered graph entity.

Appendix figures & tables2 assets

Supplementary material from the paper’s appendix.

Appendix

Dataset	Embedding dimension	Hidden dimensions	R-GCN bases	NN decoder units
AIFB	32	$[64,64,64]$	40	-
CoDEx	32	$[64,128]$	-	$[32,8,1]$
BioKG	128	$[256,256]$	-	$[64,32,8,1]$

Appendix

Table 4: Hyperparameters for the different datasets.

		ComplEx		R-GCN / GraphSAGE
Dataset	Metric	Baseline	SC-Edges	Baseline	SC-Edges	SemLoss
AIFB	Hits@1	0.2389 \smaller $\pm$ 0.0156	0.2764 \smaller $\pm$ 0.0113	0.3478 \smaller $\pm$ 0.0227	0.3565 \smaller $\pm$ 0.0289	0.3861 § \smaller $\pm$ 0.0157
	Hits@3	0.5041 \smaller $\pm$ 0.0103	0.5303 \smaller $\pm$ 0.0106	0.5334 \smaller $\pm$ 0.0311	0.5454 \smaller $\pm$ 0.0275	0.5671 ∗†‡ \smaller $\pm$ 0.0190
	Hits@10	0.7327 \smaller $\pm$ 0.0113	0.7334 \smaller $\pm$ 0.0042	0.7409 \smaller $\pm$ 0.0306	0.7558 \smaller $\pm$ 0.0289	0.7687 ∗†‡ \smaller $\pm$ 0.0127
CoDEx	Hits@1	0.3020 \smaller $\pm$ 0.0019	0.2971 \smaller $\pm$ 0.0020	0.3266 \smaller $\pm$ 0.0078	0.3310 \smaller $\pm$ 0.0082	0.3380 § \smaller $\pm$ 0.0030
	Hits@3	0.4111 \smaller $\pm$ 0.0008	0.4085 \smaller $\pm$ 0.0016	0.4423 \smaller $\pm$ 0.0066	0.4448 \smaller $\pm$ 0.0070	0.4478 †‡ \smaller $\pm$ 0.0045

Appendix

Table 5: Link prediction performance (Hits@K). Results are reported as mean

\pm

standard deviation over 10 runs.

\lx@sectionsign

indicates significant improvement for the model trained with the semantic loss over all baseline models.

\dagger

and

\ddagger

indicate significant improvements compared to the ComplEx and ComplEx with subClassOf -edges respectively. * indicates significant improvement compared to the GNN baseline. Improvements are tested with Welch’s two-sided t-test (

p<0.05

KG Completion

Knowledge graph completion (KGC) predicts missing entities, relations, or facts from structured graphs and supporting evidence.

See all 76 papers

Graph Neural Network based Hierarchy-Aware Embeddings of Knowledge Graphs: Applications to Yeast Phenotype Prediction
May 5, 2026Filip Kronström, Alexander H. Gower, Daniel Brunnsåker +2Graph Neural Networks Representation Learning
Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Sweden · Department of Life Sciences, Chalmers University of Technology, Sweden · Department of Industrial Biotechnology, KTH Royal Institute of Technology, Sweden +1
Leveraging Graph Structure in Seq2Seq Models for Knowledge Graph Link Prediction
May 18, 2026Luu Huu Phuc, Ratan Bahadur Thapa, Mojtaba Nayyeri +3Sequence Modeling KG Completion
Analytic Computing, KI, University of Stuttgart, Stuttgart, Germany · Bosch Center for Artificial Intelligence, Stuttgart, Germany · WAIS, University of Southampton, United Kingdom
Half a Link can Be Enough to Predict a Whole Link: Understanding Generalization in Knowledge Graph Foundation Models
Jun 16, 2026Cosimo Gregucci, Obaidah Theeb, Daniel Hernandez +2Cross-Domain Generalization Graph Foundation Models
Institute for AI, University of Stuttgart · University of Edinburgh · University of Southampton

May 5, 2026J/K move · Enter open · S save

Graph Neural Network based Hierarchy-Aware Embeddings of Knowledge Graphs: Applications to Yeast Phenotype Prediction

Filip Kronström, Alexander H. Gower, Daniel Brunnsåker, Ievgeniia A. Tiukova+1

Department of Computer Science and Engineering, Chalmers University of Technology and University of Gothenburg, Sweden · Department of Life Sciences, Chalmers University of Technology, Sweden · Department of Industrial Biotechnology, KTH Royal Institute of Technology, Sweden+1

Graph Neural Networks Representation Learning Graph Representation Learning Hierarchical Representation Learning KG Embedding Knowledge Graphs+1

Open paper AskPDF arXiv

We present a method for finding hierarchy-aware embeddings of knowledge graphs (KGs) using graph neural networks (GNNs) enriched with a semantic loss derived from underlying ontologies. This method yields embeddings that better reflect domain knowledge. To demonstrate their utility, we predict and interpret the effects of gene deletions in the yeast Saccharomyces cerevisiae and learn box embeddings for KGs in the absence of a prediction task. We further show how box embeddings can serve as the basis for evaluating KG revisions. Our yeast KG is constructed from community databases and ontology terms. Low-dimensional box embeddings combined with GNNs are used to predict cell growth for double gene knockouts. Over 10-fold cross validation, these predictions have a mean

R^2

~~score~~of~0.360, significantly higher than baseline comparisons, demonstrating that high-level qualitative knowledge is informative about experimental outcomes. Incorporating semantic loss terms in the training of the models improves their predictive performance (

R^2

=0.377) by aligning embeddings with ontology structure. This shows that class hierarchies from ontologies can be exploited for quantitative prediction. We also test the trained models on triple gene knockouts, showing they generalise to data beyond those seen in training. Additionally, by identifying co-occurring relations in the yeast KG important for the cell-growth predictions, we construct hypotheses about interacting traits in yeast. A biological experiment validates one such finding, revealing an association between inositol utilisation and osmotic stress resistance, highlighting the model's potential to guide biological discovery.

Abstract

KG Completion

Explore similar work

Graph Neural Network based Hierarchy-Aware Embeddings of Knowledge Graphs: Applications to Yeast Phenotype Prediction

Leveraging Graph Structure in Seq2Seq Models for Knowledge Graph Link Prediction

Half a Link can Be Enough to Predict a Whole Link: Understanding Generalization in Knowledge Graph Foundation Models