Article: Cross-lingual sentiment analysis for low-resource languages via semantic alignment and transfer learning Journal: International Journal of Information and Communication Technology (IJICT) 2025 Vol.26 No.33 pp.1 - 17 Abstract: This study addresses the challenges of sentiment analysis for low-resource languages by proposing a cross-lingual transfer augmentation method (CTAM) that integrates semantic alignment and adversarial training. Leveraging a fusion module (VXLM) combining VecMap and XLM-R, the framework dynamically aligns static and contextualised embeddings across English and Burmese languages. A dual-path attention supervision mechanism transfers sentiment knowledge from high-resource English to low-resource Burmese while mitigating cultural and structural disparities. Experiments on Myanmar social media data demonstrate state-of-the-art F1-scores, outperforming baseline models. Ablation studies validate the efficacy of the VXLM module and attention-based knowledge distillation. This work advances multilingual NLP by providing a scalable solution for low-resource language processing. Inderscience Publishers - linking academia, business and industry through research

Title: Cross-lingual sentiment analysis for low-resource languages via semantic alignment and transfer learning

Authors: Pan Zhou; Li Zhao

Addresses: School of Foreign Languages, Guangzhou Huashang College, Guangzhou, Guangdong, 511300, China ' School of Automation and Software Engineering, Shanxi University, Taiyuan, Shanxi, 030006, China

Abstract: This study addresses the challenges of sentiment analysis for low-resource languages by proposing a cross-lingual transfer augmentation method (CTAM) that integrates semantic alignment and adversarial training. Leveraging a fusion module (VXLM) combining VecMap and XLM-R, the framework dynamically aligns static and contextualised embeddings across English and Burmese languages. A dual-path attention supervision mechanism transfers sentiment knowledge from high-resource English to low-resource Burmese while mitigating cultural and structural disparities. Experiments on Myanmar social media data demonstrate state-of-the-art F1-scores, outperforming baseline models. Ablation studies validate the efficacy of the VXLM module and attention-based knowledge distillation. This work advances multilingual NLP by providing a scalable solution for low-resource language processing.

Keywords: sentiment analysis; multilingual NLP; VecMap and XLM-R; cross-lingual transfer augmentation.

DOI: 10.1504/IJICT.2025.148657

International Journal of Information and Communication Technology, 2025 Vol.26 No.33, pp.1 - 17

Received: 14 May 2025
Accepted: 26 Jun 2025
Published online: 17 Sep 2025 *

Title: Cross-lingual sentiment analysis for low-resource languages via semantic alignment and transfer learning

Keep up-to-date