基于知识库检索反馈的期货市场智能合规方法研究

Research on the Intelligent Compliance Methods in the Futures Market Based on Knowledge Base Retrieval Feedback

摘要

随着人工智能在高合规性领域的深入应用,如何在保障模型表达能够满足严格法规约束,成为大语言模型面临的核心挑战.针对现有方法存在的标注成本高、难以适应动态监管场景及偏好与规则冲突等问题,本文提出一种基于合规知识库反馈的强化学习微调框架(RLKBF,Reinforcement Learning from Knowledge Base Feedback).该方法结合语义增强的向量表示与层次化检索机制,构建结构化法规知识库.同时,引入合规偏离惩罚项的双目标优化策略,以协调用户意图与法规规范的平衡.实验结果显示,RLKBF在回答准确率、合规性稳定性及专家评估指标上均优于主流对比模型,显著提升了模型对专业法规知识的整合与应用能力.

Abstract

As artificial intelligence continues to be applied in highly regulated domains,ensuring that large language models comply with strict legal constraints has become a central challenge.Existing approaches often face limitations such as high annotation costs,poor adaptability to evolving regulatory environments,and conflicts between user preferences and compliance requirements.To address these issues,this paper proposes a fine-tuning framework based on Reinforcement Learning from Knowledge Base Feedback(RLKBF).This method constructs a structured legal knowledge base using semantically enhanced vector representations and hierarchical retrieval mechanisms.Additionally,a dual-objective optimization strategy is introduced,incorporating a compliance deviation penalty to balance user intent with regulatory constraints.Experimental results demonstrate that RLKBF outperforms mainstream models in terms of response accuracy,compliance stability,and expert evaluation metrics,significantly improving the model's ability to integrate and apply domain-specific legal knowledge.

关键词

期货市场 / 大语言模型 / 模型微调 / 知识库构建 / 合规管理

Key words

futures market / large language models / model fine-tuning / knowledge base construction / compliance management

[1] Liu Z,Huang D,Huang K,et al.Finbert:a pre-trained financial language representation model for financial text mining[C]//Proceedings of the 29th International Conference on International Joint Conferences on Artificial Intelligence,2021:4513-4519.
[2] Nguyen D,Cao N,Nguyen S,et al.FinBERT:multilingual pretrained language model for financial domain[C]//14th International Conference on Knowledge and Systems Engineering,2024,doi:10.1109/KSE56063.2022.9953749.
[3] WChao P,Robey A,Dobriban E,et al.Jailbreaking black box large language models in twenty queries[C]//IEEE Conference on Secure and Trustworthy Machine Learning,2025:23-42.
[4] Zheng J,Qiu S,Shi C,et al.Towards lifelong learning of large language models:a survey[J].ACM Computing Surveys,2025,57(8):1-35.
[5] Xie Q,Han W,Chen Z,et al.Finben:a holistic financial benchmark for large language models[J].Advances in Neural Information Processing Systems,2024,37(3):95716-95743.
[6] Li Y,Wang S,Ding H,et al.Large language models in finance:a survey[C]//Proceedings of the 4th ACM International Conference on AI in Finance,2023:374-382.
[7] Lewis P,Perez E,Piktus A,et al.Retrieval-augmented generation for knowledge-intensive NLP tasks[J].Advances in Neural Information Processing Systems,2020,33(5):9459-9474.
[8] Fan W,Ding Y,Ning L,et al.A survey on rag meeting llms:towards retrieval-augmented large language models[C]//Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining,2024:6491-6501.
[9] Cm H,Das D,Ranjan R K,et al.LoKI:money laundering report generation via logical table-to-text using meta learning[C]//Proceedings of the 5th Workshop on Financial Technology and Natural Language Processing and the Second Multimodal AI for Financial Forecasting,2023:104-110.
[10] Darji A,Kheni F,Chodvadia D,et al.Enhancing financial risk analysis using rag-based large language models[C]//3rd International Conference on Automation,Computing and Renewable Systems,2024:754-760.
[11] Malandri L,Mercorio F,Mezzanzanica M,et al.RE-FIN:retrieval-based enrichment for financial data[C]//Proceedings of the 31st International Conference on Computational Linguistics:Industry Track,2025:751-759.
[12] Chen W,Su Y,Yan X,et al.KGPT:knowledge-grounded pre-training for data-to-text generation[C]//Conference on Empirical Methods in Natural Language Processing,2020:8635-8648.
[13] Han Y,Wang T.Semi-supervised clustering for financial risk analysis[J].Neural Processing Letters,2021,53(5):3561-3572.
[14] Hu L,Liu Z,Zhao Z,et al.A survey of knowledge enhanced pre-trained language models[J].IEEE Transactions on Knowledge and Data Engineering,2024,36(4):1413-1430.
[15] Loster M.Knowledge base construction with machine learning methods[D].Potsdam:Universität Potsdam,2021.
[16] Iaroshev I,Pillai R,Vaglietti L,et al.Evaluating retrieval-augmented generation models for financial report question and answering[J].Applied Sciences,2024,14(20):2076-3417.
[17] Zakka C,Shad R,Chaurasia A,et al.Almanac—retrieval-augmented language models for clinical medicine[J].New England Journal of Medicine AI,2024,1(2),doi:10.21203/rs.3.rs-2883198/v1.
[18] Lee H,Phatale S,Mansoor H,et al.RLAIF vs.RLHF:scaling reinforcement learning from human feedback with AI feedback[C]//Proceedings of the 41st International Conference on Machine Learning,2024:26874-26901.
[19] Zhang D,Zhoubian S,Hu Z,et al.Rest-mcts^*:LLM self-training via process reward guided tree search[J].Advances in Neural Information Processing Systems,2024,37(9):64735-64772.
[20] Xu W,Fang M,Yang L,et al.Enabling language representation with knowledge graph and structured semantic information[C]//International Conference on Computer Communication and Artificial Intelligence,2021:91-96.
[21] Chen B,Bertozzi A L.AutoKG:efficient automated knowledge graph generation for language models[C]//IEEE International Conference on Big Data,2023:3117-3126.
[22] Yu Y,Yan Y,Jin Y.Structural knowledge:from brain to artificial intelligence[J].Artificial Intelligence Review,2025,58(9):1-39.

基金资助

国家自然科学基金项目(71971066)资助.