基于大语言模型RAG技术的金融交易反欺诈识别方法研究

Research on Anti-Fraud Recognition Method for Financial Transactions Based on Large Language Model RAG Technology

摘要: 本文深入探讨了大型语言模型(Large Language Models, LLMs)在金融交易反欺诈识别中的应用，特别是通过引入检索增强生成技术(Retrieval-Augmented Generation, RAG)来优化模型的性能和准确性。RAG技术通过在生成答案前从广泛的向量数据库中检索相关信息，引导模型生成更精确和可靠的答案。本文评估了RAG技术在金融交易反欺诈识别任务中的效果，并提出了相应的方法论和实验结果。实验表明，RAG技术能显著提升模型在识别金融欺诈行为上的能力，同时保持模型的通用性和适应性。

Abstract: This paper provides an in-depth discussion on the application of Large Language Models (LLMs) in anti-fraud recognition of financial transactions, in particular, optimizing the performance and accuracy of the models by introducing the Retrieval-Augmented Generation (RAG) technique, which guides the models to generate more accurate and reliable answers by retrieving relevant information from a wide range of databases, the RAG technique guides the model to generate more accurate and reliable answers. This paper evaluates the effectiveness of the RAG technique in the anti-fraud recognition task of financial transactions and presents the corresponding methodology and experimental results. The experiments show that the RAG technique significantly improves the model’s ability in recognizing financial frauds while maintaining the generality and adaptability of the model.