Article: Transformer-based cross-cultural intelligent translation system for international communication Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.27 pp.60 - 83 Abstract: This study proposes a Transformer-based cross-cultural intelligent translation system to enhance international communication. By integrating attention mechanisms and large-scale multilingual datasets encompassing 47,850 samples across seven languages from 35 countries, the model achieves 97.3% accuracy in predicting language competency while ensuring contextual fluency and cultural adaptability. The approach outperforms traditional and BERT-based methods, offering a scalable solution for multilingual, multicultural contexts. Language is a vital bridge for cross-cultural communication, especially in global collaborations. However, traditional translation systems struggle with contextual accuracy and cultural inclusivity. Previous studies have explored neural machine translation enhancements, such as GANs, BiLSTM generators, and syntax-aware methods. While effective, these approaches often face limitations in low-resource languages and cultural adaptability. A hybrid deep learning framework combining Transformer architecture and attention mechanisms was developed. The proposed model achieved 97.3% accuracy, 96.8% precision, 95.6% recall, and 96.4% F1-score. These results outperform state-of-the-art baselines, demonstrating superior performance in cross-cultural translation. Inderscience Publishers - linking academia, business and industry through research

Title: Transformer-based cross-cultural intelligent translation system for international communication

Authors: Sha Liu

Addresses: School of International Culture and Communication, Beijing City University, Beijing, 100191, China

Abstract: This study proposes a Transformer-based cross-cultural intelligent translation system to enhance international communication. By integrating attention mechanisms and large-scale multilingual datasets encompassing 47,850 samples across seven languages from 35 countries, the model achieves 97.3% accuracy in predicting language competency while ensuring contextual fluency and cultural adaptability. The approach outperforms traditional and BERT-based methods, offering a scalable solution for multilingual, multicultural contexts. Language is a vital bridge for cross-cultural communication, especially in global collaborations. However, traditional translation systems struggle with contextual accuracy and cultural inclusivity. Previous studies have explored neural machine translation enhancements, such as GANs, BiLSTM generators, and syntax-aware methods. While effective, these approaches often face limitations in low-resource languages and cultural adaptability. A hybrid deep learning framework combining Transformer architecture and attention mechanisms was developed. The proposed model achieved 97.3% accuracy, 96.8% precision, 95.6% recall, and 96.4% F1-score. These results outperform state-of-the-art baselines, demonstrating superior performance in cross-cultural translation.

Keywords: transformer model; cross-cultural communication; neural machine translation; NMT; natural language processing; NLP; attention mechanism; legislative impact assessment; LIA.

DOI: 10.1504/IJICT.2026.152534

International Journal of Information and Communication Technology, 2026 Vol.27 No.27, pp.60 - 83

Received: 04 Sep 2025
Accepted: 30 Nov 2025
Published online: 25 Mar 2026 *

Title: Transformer-based cross-cultural intelligent translation system for international communication

Keep up-to-date