Multimodal entity alignment improves with better missing data handling
Silence and Noise: Resolving Suppression and Overvaluation in Multimodal Entity Alignment
Machine Learning
Summary
When combining knowledge from different sources, computers often struggle because some parts of the data are missing or the connections between data points don’t match well. The authors show that these problems interact in ways earlier methods overlooked, which makes alignment less accurate. They propose a new approach called ALMEA that fills in missing details and balances the different types of information to improve alignment. Their method works better than existing ones, especially when only limited labeled data is available.
What this means in practice
- •For knowledge graph engineers: Improve the integration of heterogeneous knowledge graphs by reducing errors caused by missing or mismatched modality details during entity alignment.
- •For data integration teams: Enhance cross-database entity matching accuracy by synthesizing missing data and balancing structural inconsistencies in multimodal inputs.
Authors
Xizhe Zhang, Jingfeng Zhang, Meng-Fen Chiang
Abstract
Multimodal entity alignment (MMEA) is fundamental to integrating heterogeneous knowledge graphs (KGs), yet existing paradigms remain critically vulnerable to incomplete modalities and cross-graph structural asymmetries. This paper exposes a fundamental oversight in current MMEA research: treating feature missingness and topological discrepancies as isolated preprocessing or denoising challenges, which overlooks their destructive coupling during multimodal fusion. We systematically formulate and evaluate two newly identified failure modes: 1) Detail Missing (DM)-induced overvaluation, where the absence of local modality details causes models to overemphasize uninformative, blurred representations; and 2) Context Imbalance (CI)-induced suppression, where topological discrepancies across KGs allow dense structural contexts to overwhelm weaker but highly discriminative modal signals, triggering high-level semantic drift. To break this cycle of degradation, we introduce ALMEA, a generative and calibration framework that establishes a bidirectional feedback loop between low-level feature reconstruction and high-level latent space alignment. Specifically, Cross-Modality Imputation (CMI) synthesizes missing modality details to mitigate overvaluation, while Latent Semantic Calibration (LSC) aligns cross-graph joint embeddings to suppress topology-driven semantic drift. Under sparse supervision, we develop Controlled Iterative Selection (CIS) to adaptively acquire representative and diverse pseudo-labeled pairs, optimizing latent calibration under constrained budgets. Extensive evaluations on standard benchmarks demonstrate that ALMEA consistently outperforms state-of-the-art baselines. At a 20% supervision rate, ALMEA yields substantial absolute improvements of up to 5.20% in MRR and 6.40% in Hits@1, maintaining robust gains across higher supervision thresholds.