Chain-of-Translation Prompting (CoTR): A Novel Prompting Technique for Low Resource Languages

2024-09-06

arXiv.org

AI Summary

Low-resource languages (e.g., Marathi) suffer from suboptimal performance in NLP tasks—including sentiment analysis, hate speech detection, topic classification, and text generation—due to limited annotated data and pretrained model support. To address this, we propose Chain-of-Translation Prompting (CoTR), a novel prompting paradigm that intrinsically integrates translation into the prompt structure: “low-resource language → high-resource language → task execution → optional back-translation.” Unlike conventional approaches treating translation as an external preprocessing step, CoTR embeds it end-to-end within a single prompt, enabling fully prompt-based, zero-shot or few-shot inference without model fine-tuning. Evaluated on Marathi, CoTR consistently outperforms standard prompting baselines across all tasks, with the largest gains observed in hate speech detection accuracy. Moreover, CoTR significantly enhances the quality of synthetically generated low-resource language data, facilitating downstream applications requiring high-fidelity linguistic output.

Technology Category

• Natural Language Processing: Machine Translation, Multilinguality, Cross-Lingual NLP
• Machine Learning: Large Multimodal Models (LMMs)
• Planning, Routing, and Scheduling: Planning with Language Models

Application Category

• Search and Retrieval-Augmented AI: Multilingual and cross-lingual Web search
• Web Mining and Content Analysis: Large pretrained models with web data
• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs

Abstract

This paper introduces Chain of Translation Prompting (CoTR), a novel strategy designed to enhance the performance of language models in low-resource languages. CoTR restructures prompts to first translate the input context from a low-resource language into a higher-resource language, such as English. The specified task like generation, classification, or any other NLP function is then performed on the translated text, with the option to translate the output back to the original language if needed. All these steps are specified in a single prompt. We demonstrate the effectiveness of this method through a case study on the low-resource Indic language Marathi. The CoTR strategy is applied to various tasks, including sentiment analysis, hate speech classification, subject classification and text generation, and its efficacy is showcased by comparing it with regular prompting methods. Our results underscore the potential of translation-based prompting strategies to significantly improve multilingual LLM performance in low-resource languages, offering valuable insights for future research and applications. We specifically see the highest accuracy improvements with the hate speech detection task. The technique also has the potential to enhance the quality of synthetic data generation for underrepresented languages using LLMs.

Problem

• Low-resource Languages
• Language Model Performance
• NLP Task Accuracy

Innovation

• CoTR
• Translation Chain Prompting
• Low-resource Language Support