CRAT: A Multi-Agent Framework for Causality-Enhanced Reflective and Retrieval-Augmented Translation with Large Language Models

TL;DR

CRAT is a novel multi-agent translation framework that leverages RAG and causality-enhanced self-reflection to address challenges in large language models, and significantly improves translation accuracy, particularly in handling context-sensitive terms and emerging vocabulary.

Abstract

Large language models (LLMs) have shown great promise in machine translation, but they still struggle with contextually dependent terms, such as new or domain-specific words. This leads to inconsistencies and errors that are difficult to address. Existing solutions often depend on manual identification of such terms, which is impractical given the complexity and evolving nature of language. While Retrieval-Augmented Generation (RAG) could provide some assistance, its application to translation is limited by issues such as hallucinations from information overload. In this paper, we propose CRAT, a novel multi-agent translation framework that leverages RAG and causality-enhanced self-reflection to address these challenges. This framework consists of several specialized agents: the Unknown Terms Identification agent detects unknown terms within the context, the Knowledge Graph (KG) Constructor agent extracts relevant internal knowledge about these terms and retrieves bilingual information from external sources, the Causality-enhanced Judge agent validates the accuracy of the information, and the Translator agent incorporates the refined information into the final output. This automated process allows for more precise and consistent handling of key terms during translation. Our results show that CRAT significantly improves translation accuracy, particularly in handling context-sensitive terms and emerging vocabulary.

Figures (2)

• Figure 1: Demonstration of the contextual ambiguity LLMs face during English-Chinese translation. In the top example, "bank" and "Scotia" are interpreted within a financial context, while in the bottom example, they shift to a geographical meaning. This illustrates the necessity for combining both internal knowledge and external, domain-specific information to accurately reason about context-dependent terms.
• Figure 2: Our proposed multi-agent translation framework (i.e., CRAT) for improving LLM translation accuracy, especially for unknown terms. The Unknown Terms Detector Agent identifies terms with low confidence or ambiguity, such as polysemes or new terms. The Knowledge Graph Constructor Agent then builds a TransKG based on both internal and external sources to collect the necessary information. The Causality-enhanced Judge Agent evaluates the appropriateness of term meanings by considering causal invariance in context. Finally, the Translator Agent uses accurate meanings to generate translations.