Search papers, labs, and topics across Lattice.
This paper addresses the challenge of imbalanced pseudo-seed coverage in unsupervised multimodal entity alignment (MMEA) by proposing PSQE, a method to enhance pseudo-seed quality. PSQE leverages multimodal information and clustering-resampling to improve both the precision and graph coverage balance of pseudo-seeds. Theoretical analysis demonstrates how pseudo-seeds impact attraction and repulsion terms in contrastive learning, and how imbalanced coverage biases models towards high-density regions.
Unbalanced pseudo-seeds in multimodal entity alignment cause models to favor high-density knowledge graph regions, but a new method corrects this imbalance and boosts performance.
Multimodal Entity Alignment (MMEA) aims to identify equivalent entities across different data modalities, enabling structural data integration that in turn improves the performance of various large language model applications. To lift the requirement of labeled seed pairs that are difficult to obtain, recent methods shifted to an unsupervised paradigm using pseudo-alignment seeds. However, unsupervised entity alignment in multimodal settings remains underexplored, mainly because the incorporation of multimodal information often results in imbalanced coverage of pseudo-seeds within the knowledge graph. To overcome this, we propose PSQE (Pseudo-Seed Quality Enhancement) to improve the precision and graph coverage balance of pseudo seeds via multimodal information and clustering-resampling. Theoretical analysis reveals the impact of pseudo seeds on existing contrastive learning-based MMEA models. In particular, pseudo seeds can influence the attraction and the repulsion terms in contrastive learning at once, whereas imbalanced graph coverage causes models to prioritize high-density regions, thereby weakening their learning capability for entities in sparse regions. Experimental results validate our theoretical findings and show that PSQE as a plug-and-play module can improve the performance of baselines by considerable margins.
