Article: The design of multi-modal intelligent vocabulary memory system for fragmented learning Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.40 pp.75 - 99 Abstract: This study addresses the limitations of traditional vocabulary memorisation, such as lack of context, low efficiency, and poor adaptability to fragmented learning. It proposes a vocabulary learning model integrating constrained translation and image-text matching. Using a transformer-based translation model with positional encoding and grid beam search, the approach enables contextualised micro-learning. A contrastive learning image-text matching model builds multimodal 'vocabulary-scene' associations to reinforce memory. An integrated vocabulary learning system is developed and evaluated in terms of objectives and design requirements. Results show that the proposed model outperforms comparative methods across multiple metrics, achieving the highest scores in delayed memory tests (88.1%) and vocabulary application (8.5/10), with a retention rate (95.2%) far exceeding other models. It also enhances learning engagement, scoring 4.4 points in user interest. This method offers a new technical and theoretical foundation for designing intelligent, personalised vocabulary learning systems, particularly suitable for mobile fragmented learning scenarios. Inderscience Publishers - linking academia, business and industry through research

Title: The design of multi-modal intelligent vocabulary memory system for fragmented learning

Authors: Fen Guo

Addresses: School of Foreign Languages, Zhongyuan Institute of Science and Technology, Zhengzhou, 451400, China

Abstract: This study addresses the limitations of traditional vocabulary memorisation, such as lack of context, low efficiency, and poor adaptability to fragmented learning. It proposes a vocabulary learning model integrating constrained translation and image-text matching. Using a transformer-based translation model with positional encoding and grid beam search, the approach enables contextualised micro-learning. A contrastive learning image-text matching model builds multimodal 'vocabulary-scene' associations to reinforce memory. An integrated vocabulary learning system is developed and evaluated in terms of objectives and design requirements. Results show that the proposed model outperforms comparative methods across multiple metrics, achieving the highest scores in delayed memory tests (88.1%) and vocabulary application (8.5/10), with a retention rate (95.2%) far exceeding other models. It also enhances learning engagement, scoring 4.4 points in user interest. This method offers a new technical and theoretical foundation for designing intelligent, personalised vocabulary learning systems, particularly suitable for mobile fragmented learning scenarios.

Keywords: vocabulary memory; transformer model; image-text matching technology; contrastive learning; learning strategy; constrained translation.

DOI: 10.1504/IJICT.2026.153315

International Journal of Information and Communication Technology, 2026 Vol.27 No.40, pp.75 - 99

Received: 03 Nov 2025
Accepted: 08 Dec 2025
Published online: 01 May 2026 *

Title: The design of multi-modal intelligent vocabulary memory system for fragmented learning

Keep up-to-date