SASA: Semantic-Aware Contrastive Learning Framework with Separated Attention for Triple Classification

TL;DR

This work targets triple classification in knowledge graphs by addressing semantic interaction gaps and limited semantic learning from a single objective. It introduces SASA, a framework combining a dual-tower encoding with separated attention and a hierarchical semantic-aware contrastive learning module that includes local-level dropout-based positives and global-level hard negatives. The approach yields state-of-the-art results on FB15k-237 and YAGO3-10, with notable gains over baselines and robust ablations demonstrating the contribution of each component. The findings suggest that separating component representations and reinforcing them with multi-level contrastive supervision enhances fine-grained semantic discrimination in KGs, enabling more reliable TC and potentially benefiting downstream KG applications.

Abstract

Knowledge Graphs~(KGs) often suffer from unreliable knowledge, which restricts their utility. Triple Classification~(TC) aims to determine the validity of triples from KGs. Recently, text-based methods learn entity and relation representations from natural language descriptions, significantly improving the generalization capabilities of TC models and setting new benchmarks in performance. However, there are still two critical challenges. First, existing methods often ignore the effective semantic interaction among different KG components. Second, most approaches adopt single binary classification training objective, leading to insufficient semantic representation learning. To address these challenges, we propose \textbf{SASA}, a novel framework designed to enhance TC models via separated attention mechanism and semantic-aware contrastive learning~(CL). Specifically, we first propose separated attention mechanism to encode triples into decoupled contextual representations and then fuse them through a more effective interactive way. Then, we introduce semantic-aware hierarchical CL as auxiliary training objective to guide models in improving their discriminative capabilities and achieving sufficient semantic learning, considering both local level and global level CL. Experimental results across two benchmark datasets demonstrate that SASA significantly outperforms state-of-the-art methods. In terms of accuracy, we advance the state-of-the-art by +5.9\% on FB15k-237 and +3.4\% on YAGO3-10.

Figures (3)

• Figure 1: The overall framework of our SASA.
• Figure 2: Effect of negative sample quantity.
• Figure 3: Visualization of entity embeddings.