赵克清,王一群,陈雯柏

• 1:北京信息科技大学自动化学院

摘要(Abstract)：

随着农业决策的日益复杂，现有的检索增强生成（retrieval augmented generation,RAG）模型在处理带有强条件依赖性的农业知识时，常常无法有效捕捉条件约束对信息检索的影响，导致检索结果与实际需求不匹配。为了解决这一问题，提出了一种条件感知融合知识检索（conditionaware fusion knowledge retrieval, CAFKR）方法，通过优化检索过程提升RAG系统的检索精准度与领域适应性。CAFKR通过引入条件识别框架精准解析农业知识中的条件约束关系，并结合多路径检索机制与自适应信息融合策略构建条件与信息的精准映射，有效增强了对农业领域条件依赖性的处理能力。实验结果表明，CAFKR在Recall@5指标上达到了0.910 1，显著优于传统的检索模型。该方法为后续RAG系统的开发提供了可靠基础，能够在农业等复杂领域中提供更加精准和高效的知识检索。

关键词(KeyWords)： 精准农业;条件感知融合;多路径检索;检索增强生成

Abstract:

Keywords:

基金项目(Foundation): 科技部2030新一代人工智能重大专项（2021ZD0113603）;; 国家自然科学基金项目（62276028）;国家自然科学基金重大研究计划（92267110）

作者(Author): 赵克清,王一群,陈雯柏

DOI: 10.16508/j.cnki.11-5866/n.2025.06.005

参考文献(References)：

• [1]方松，姜丽华，曹景军，等. AI for Science在农业领域的应用研究[J].中国农业科技导报，2024,26(10):1-10.FANG S,JIANG L H,CAO J J,et al. Research on the application of AI for Science in agricultural field[J]. Journal of Agricultural Science and Technology,2024,26(10):1-10.(in Chinese)
• [2]赵春江.农业知识智能服务技术综述[J].智慧农业（中英文），2023,5(2):126-148.ZHAO C J. Agricultural knowledge intelligent service technology:a review[J]. Smart Agriculture,2023,5(2):126-148.(in Chinese)
• [3]王元胜，吴华瑞，赵春江.农业知识驱动服务技术革新综述与前沿[J].农业工程学报，2024,40(7):1-16.WANG Y S,WU H R,ZHAO C J. Agricultural knowledge driven service technology innovation:overview and frontiers[J].Transactions of the Chinese Society of Agricultural Engineering,2024,40(7):1-16.(in Chinese)
• [4]WU L K,ZHENG Z,QIU Z P,et al. A survey on large language models for recommendation[J]. World Wide Web,2024,27(5):60.
• [5]姜京池，闫莲，刘劼.基于精准知识筛选及知识协同生成的农业大语言模型[J].智慧农业（中英文），2025,7(1):20-32.JIANG J C,YAN L,LIU J. Agricultural large language model based on precise knowledge retrieval and knowledge collaborative generation[J]. Smart Agriculture,2025,7(1):20-32.(in Chinese)
• [6]刘雪颖，云静，李博，等.基于大型语言模型的检索增强生成综述[J].计算机工程与应用，2025,61(13):1-25.LIU X Y,YUN J,LI B,et al. Survey of retrieval-augmented generation based on large language models[J]. Computer Engineering and Applications,2025,61(13):1-25.(in Chinese)
• [7]WAN Y W,CHEN Z Y,LIU Y,et al. Empowering LLMs by hybrid retrieval-augmented generation for domain-centric Q&A in smart manufacturing[J]. Advanced Engineering Informatics,2025,65(Part B):103212.
• [8]VIZNIUK A,DIACHENKO G,LAKTIONOV I,et al. A comprehensive survey of retrieval-augmented large language models for decision making in agriculture:unsolved problems and research opportunities[J]. Journal of Artificial Intelligence and Soft Computing Research,2024,15(2):115-146.
• [9]GUO M L,JIA X P,HUANG J K,et al. Farmer field school and farmer knowledge acquisition in rice production:experimental evaluation in China[J]. Agriculture, Ecosystems&Environment,2015,209:100-107.
• [10]GHANNEM A,BEN AISSA I,MAJDOUB R. Effects of regulated deficit irrigation applied at different growth stages of greenhouse grown tomato on substrate moisture,yield,fruit quality,and physiological traits[J]. Environmental Science and Pollution Research International,2021,28(34):46553-46564.
• [11]YU H,GAN A R,ZHANG K,et al. Evaluation of retrievalaugmented generation:a survey[C]//12th CCF Conference on BigData. Berlin, Germany:Springer Science and Business Media Deutschland GmbH,2025, 2301:102-120.
• [12]GUO D Y,YANG D J,ZHANG H W,et al. DeepSeek-R1incentivizes reasoning in LLMs through reinforcement learning[J]. Nature,2025,645(8081):633-638.
• [13]ROBERTSON S,ZARAGOZA H. The probabilistic relevance framework:BM25 and beyond[J]. Foundations and Trends in Information Retrieval,2009,3(4):333-389.
• [14]XU J Q,ZHU J J,BA Y H. M3-RAG:unified multimodal and multilingual retrieval-augmented generation[C]//2025 6th International Conference on Computer Engineering and Application(ICCEA). New York, USA:IEEE,2025:1380-1383.
• [15]PANG T Q,TAN K H,YAO Y J,et al. REMED:retrievalaugmented medical document query responding with embedding fine-tuning[C]//2024 International Joint Conference on Neural Networks(IJCNN). New York, USA:IEEE,2024:1-8.
• [16]OMOTSOS A,SHOPEJU H,JOSHUA A M,et al. Improving BGE-M3 multilingual dense embeddings for Nigerian low resource languages[C]//Proceedings of the 9th Widening NLP Workshop. Stroudsburg, PA, USA:Association for Computational Linguistics,2025:224-229.
• [17]MANDIKAL P,MOONEY R. Sparse meets dense:a hybrid approach to enhance scientific document retrieval[EB/OL].(2024-01-08)[2025-10-20]. http：//arxiv. org/abs/2401. 04055.
• [18]PATEL Y,TOLIAS G,MATAS J. Recall@k surrogate loss with large batches and similarity mixup[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition(CVPR). Los Alamitos, CA ,USA:IEEE Computer Society,2022:7492-7501.
• [19]YU L,ZHOU G,ZHANG C X,et al. RankMBPR:rank-aware mutual Bayesian personalized ranking for item recommendation[C]//Web-Age Information Management. Berlin, Germany,Springer-Verlag Berlin,2016:244-256.
• [20]NAI R,SULIS E,FATIMA I,et al. Large language models and recommendation systems:a proof-of-concept study on public procurements[C]//Natural Language Processing and Information System. Berlin, Germany:Springer Science and Business Media Deutschland GmbH,2024, 14763:280-290.
• [21]TANEJA K, VASHISHTHA J, RATNOO S. Efficient deep pretrained sentence embedding model for similarity search[J].International Journal of Computer Information Systems&Industrial Management Applications,2023,15:605-615.