aiqbench analysis
This research introduces KMAS, a method that improves Knowledge Graph Foundation Models by dynamically adjusting the use of hard negative samples during training. By leveraging updated relation embeddings to construct higher-quality negatives and adaptively varying their ratio over time, the approach significantly boosts model performance across 44 datasets without incurring extra computational costs. The work addresses a critical bottleneck in zero-shot knowledge graph completion where standard random sampling fails to provide sufficient supervision for unseen relational vocabularies.
Boosting Knowledge Graph Foundation Models via Enhanced Negative Sampling
arXiv:2605.27023v3 Announce Type: replace Abstract: Knowledge graphs (KGs) have become the core backbone of numerous downstream tasks such as question answering and recommender systems. However, despite all this, KGs are often very incomplete. To perform zero-shot knowledge graph completion in unseen KGs, which have different relational vocabularies from those used for pre-training, KG foundation models (KGFMs) receive a wide range of attention. Existing KGFMs often perform training using random negative triples, which are constructed by replacing the head or tail entity of a positive triple with a random entity. However, these negative triples are often constructed with limited quality, providing weak supervision for KGFM training. In this paper, we propose a simple yet effective adaptive negative sampling approach, KMAS, to enhance existing KGFMs. KMAS constructs hard negative triples through the updated relation embeddings generated from the existing KGFM's relation encoder. To further adaptively align with the evolving capability of the KGFM during the training process, KMAS adjusts the ratio of hard negative triples dynamically throughout the whole training process: after a warmup phrase, it increases the ratio linearly and then decreases linearly. Extensive experiments are conducted over 44 data sets. Experimental results demonstrate that our proposed negative sampling method can enhance many SOTA KGFMs without requiring excessive additional time or memory consumption.
Abstract:Knowledge graphs (KGs) have become the core backbone of numerous downstream tasks such as question answering and recommender systems. However, despite all this, KGs are often very incomplete. To perform zero-shot knowledge graph completion in unseen KGs, which have different relational vocabularies from those used for pre-training, KG foundation models (KGFMs) receive a wide range of attention. Existing KGFMs often perform training using random negative triples, which are constructed by replacing the head or tail entity of a positive triple with a random entity. However, these negative triples are often constructed with limited quality, providing weak supervision for KGFM training. In this paper, we propose a simple yet effective adaptive negative sampling approach, KMAS, to enhance existing KGFMs. KMAS constructs hard negative triples through the updated relation embeddings generated from the existing KGFM's relation encoder. To further adaptively align with the evolving capability of the KGFM during the training process, KMAS adjusts the ratio of hard negative triples dynamically throughout the whole training process: after a warmup phrase, it increases the ratio linearly and then decreases linearly. Extensive experiments are conducted over 44 data sets. Experimental results demonstrate that our proposed negative sampling method can enhance many SOTA KGFMs without requiring excessive additional time or memory consumption.
Artificial Intelligence (cs.AI)
arXiv:2605.27023 [cs.AI]
(or arXiv:2605.27023v3 [cs.AI] for this version)
Full text under the publisher's open licence · 218 words
Related coverage
Boosting Knowledge Graph Foundation Models via Enhanced Negative Sampling
arXiv · Jun 16, 2026
Boosting Knowledge Graph Foundation Models via Enhanced Negative Sampling
arXiv · Sep 30, 2026
KGPFN: Unlocking the Potential of Knowledge Graph Foundation Model via In-Context Learning
arXiv · Oct 6, 2026
Aligning Biomedical Texts and Knowledge Graphs: A Systematic Comparison of Lightweight Alignment Strategies
arXiv · Aug 25, 2026