Graph-Assisted Culturally Adaptable Idiomatic Translation for Indic Languages

2025-05-28

AI Summary

Idiom translation faces core challenges including strong cultural dependency, one-to-many lexical mappings, and low-accuracy in low-resource settings—issues inadequately addressed by static knowledge graphs or conventional prompting methods due to their inability to model dynamic cross-cultural semantic mappings. This paper proposes IdiomCE, a culture-aware adaptive graph neural network framework that constructs a culture-enhanced idiom graph and incorporates a lightweight adaptation mechanism to enable generalizable reasoning over both seen and unseen idiom nodes. IdiomCE eliminates reliance on handcrafted rules and strong supervision, and introduces reference-free evaluation metrics to improve cross-cultural paraphrasing quality from English to multiple Indian languages. Experiments demonstrate IdiomCE’s superior performance under resource-constrained conditions—particularly with small-scale models—and it achieves state-of-the-art results across multiple benchmark datasets.

Technology Category

• Natural Language Processing: Lexical Semantics and Morphology
• Knowledge Representation and Reasoning: Common-Sense Reasoning
• Cognitive Modeling & Cognitive Systems: Conceptual Inference and Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

Translating multi-word expressions (MWEs) and idioms requires a deep understanding of the cultural nuances of both the source and target languages. This challenge is further amplified by the one-to-many nature of idiomatic translations, where a single source idiom can have multiple target-language equivalents depending on cultural references and contextual variations. Traditional static knowledge graphs (KGs) and prompt-based approaches struggle to capture these complex relationships, often leading to suboptimal translations. To address this, we propose IdiomCE, an adaptive graph neural network (GNN) based methodology that learns intricate mappings between idiomatic expressions, effectively generalizing to both seen and unseen nodes during training. Our proposed method enhances translation quality even in resource-constrained settings, facilitating improved idiomatic translation in smaller models. We evaluate our approach on multiple idiomatic translation datasets using reference-less metrics, demonstrating significant improvements in translating idioms from English to various Indian languages.

Problem

• Translating idioms requires understanding cultural nuances in source and target languages
• Traditional methods fail to capture complex one-to-many idiomatic translation relationships
• Existing approaches struggle with resource-constrained settings for Indic languages

Innovation

• Adaptive GNN for idiomatic translation
• Handles cultural nuances in Indic languages
• Improves small model translation quality