Research on Cross-Modal Semantic Alignment Methods for Low-Resource Languages

Zhizhi Yu

European Journal of AI, Computing & Informatics, 2025, vol. 1, issue 3, 60-68

Abstract: This study addresses the challenge of cross-modal semantic alignment in low-resource languages, a critical problem for enabling inclusive and equitable AI-driven multimodal applications. We propose a novel framework that synergistically integrates multi-level textual embeddings, visual Transformer modeling, and the construction of a unified cross-modal projection space. To enhance alignment quality, the approach incorporates advanced mechanisms including contrastive learning, distributed semantic constraints, and fine-grained local alignment strategies. Furthermore, to mitigate data scarcity inherent in low-resource settings, we leverage transfer enhancement techniques such as cross-lingual knowledge distillation, pseudo-pair augmentation, and multi-task training. Comprehensive experiments on the FLORES-200 dataset demonstrate that our method consistently surpasses state-of-the-art models such as CLIP and ALIGN across multiple metrics. Specifically, significant gains are observed in Recall@1 and Mean Rank for languages including Swahili and Sinhala, underscoring the method's effectiveness, robustness, and generalizability in low-resource scenarios. These findings highlight the potential of the proposed approach for advancing cross-lingual multimodal understanding and bridging the performance gap for underrepresented languages.

Keywords: low-resource languages; cross-modal semantic alignment; contrastive learning; transfer enhancement (search for similar items in EconPapers)
Date: 2025
References: Add references at CitEc
Citations:

Downloads: (external link)
https://pinnaclepubs.com/index.php/EJACI/article/view/340/343 (application/pdf)

Related works:
This item may be available elsewhere in EconPapers: Search for items with the same title.

Export reference: BibTeX RIS (EndNote, ProCite, RefMan) HTML/Text

Persistent link: https://EconPapers.repec.org/RePEc:dba:ejacia:v:1:y:2025:i:3:p:60-68

Access Statistics for this article

More articles in European Journal of AI, Computing & Informatics from Pinnacle Academic Press
Bibliographic data for series maintained by Joseph Clark ().