LinYun：林业领域智能问答与决策支持的大型语言模型研究

赵维康; 王浩宇; 华净; 刘耀兵; 王秀娟; 康孟珍

doi:10.11959/j.issn.2096-6652.202603

Chinese

您当前的位置：

首页 >

文章列表页 >

LinYun：林业领域智能问答与决策支持的大型语言模型研究

学术论文 | 更新时间：2026-04-27

- LinYun：林业领域智能问答与决策支持的大型语言模型研究
- LinYun: a domain-specific large language model for intelligent question answering and decision support in forestry
- 智能科学与技术学报 2026年8卷第1期页码：83-93
- 作者机构：
  
  1.中国科学院自动化研究所多模态人工智能系统全国重点实验室，北京 100190
  2.中国科学院大学人工智能学院，北京 100049
  3.北京江恒数智生态科技有限公司，北京 102628
  4.澳门科技大学创新工程学院，澳门 999078
- 作者简介：
  
  [ "赵维康（2000- ），男，中国科学院自动化研究所多模态人工智能系统全国重点实验室硕士生，主要研究方向为面向林业应用的大语言模型与智能体。" ]
  [ "王浩宇（1984- ），男，中国科学院自动化研究所多模态人工智能系统全国重点实验室副研究员，主要研究方向为植物生长建模、智慧农业、编程语言及信息系统。" ]
  [ "华净（1981- ），男，中国科学院自动化研究所多模态人工智能系统全国重点实验室助理研究员，主要研究方向为植物生长建模、智慧农业、编程语言及计算机图形学。" ]
  [ "刘耀兵（1982- ），男，北京江恒数智生态科技有限公司总经理，主要研究方向为森林经理学及大模型推广应用。" ]
  [ "王秀娟（1982- ），女，中国科学院自动化研究所多模态人工智能系统全国重点实验室副研究员，主要研究方向为平行农业和植物建模。" ]
  [ "康孟珍（1975- ），女，中国科学院自动化研究所多模态人工智能系统全国重点实验室副研究员，主要研究方向为平行农业和计算植物。" ]
- 基金信息：
  
  国家自然科学基金项目(62076239)
- DOI：10.11959/j.issn.2096-6652.202603
  中图分类号： TP391.4
- 收稿：2025-09-30，
  
  修回：2026-02-02，
  
  录用：2026-03-12，
  
  纸质出版：2026-03-15
- 稿件说明：
移动端阅览
赵维康,王浩宇,华净等.LinYun：林业领域智能问答与决策支持的大型语言模型研究[J].智能科学与技术学报,2026,08(01):83-93.

Zhao Weikang,Wang Haoyu,Hua Jing,et al.LinYun: a domain-specific large language model for intelligent question answering and decision support in forestry[J].Chinese Journal of Intelligent Science and Technology,2026,08(01):83-93.
赵维康,王浩宇,华净等.LinYun：林业领域智能问答与决策支持的大型语言模型研究[J].智能科学与技术学报,2026,08(01):83-93. DOI： 10.11959/j.issn.2096-6652.202603.

Zhao Weikang,Wang Haoyu,Hua Jing,et al.LinYun: a domain-specific large language model for intelligent question answering and decision support in forestry[J].Chinese Journal of Intelligent Science and Technology,2026,08(01):83-93. DOI： 10.11959/j.issn.2096-6652.202603.

摘要

林业知识具有高度的专业性，涵盖范围广泛，并对法规及实践标准高度敏感。针对目前通用大语言模型在林业场景中存在的知识缺口、术语歧义以及事实性错误等问题，提出“数据合成-模型训练-系统化评测”一体化框架，基于通用基座模型进行领域指令微调，得到面向林业领域适配的模型林云（LinYun）。实验结果表明，LinYun在林业相关任务上显著优于同规模的通用模型，在部分任务上接近甚至超过更大规模模型的表现。

Abstract

Forestry knowledge is highly specialized and broad in scopeand is particularly sensitive to regulations and practical standards. To address the knowledge gaps

terminological ambiguities

and factual inaccuracies that general-purpose large language model (LLM) exhibit in forestry scenarios

an integrated framework of "data synthesis-model training-systematic evaluation" was proposed. Based on a general base model

domain-specific instruction fine-tuning was conducted to obtain LinYun

a domain-adapted model for forestry. Experimental results demonstrate that LinYun significantly outperforms general-purpose models of comparable scale in forestry-related tasks

and in some tasks approaches or even surpasses the performance of much larger models.

关键词

Keywords

references

Tian Y H, Gan R Y, Song Y, et al. ChiMed-GPT: a Chinese medical large language model with full training regime and better alignment to human preferences[C]//Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). Stroudsburg: ACL, 2024: 7156-7173.

Wang G Y, Yang G X, Du Z X, et al. ClinicalGPT: large language models finetuned with diverse medical data and comprehensive evaluation[PP]. V1. (2023-06-16)[2025-09-30]. arXiv: arXiv.2306.09968.

Chen J Y, Wang X D, Ji K, et al. HuatuoGPT-II, one-stage training for medical adaption of LLMs[PP]. V2. (2024-09-15)[2025-09-30]. arXiv: arXiv.2311.09774.

Zhou Z, Shi J X, Song P X, et al. LawGPT: a Chinese legal knowledge-enhanced large language model[PP]. V1. (2024-06-07)[2025-09-30]. arXiv: arXiv.2406.04614.

Huang Q Z, Tao M X, Zhang C, et al. Lawyer LLaMA technical report[PP]. V2. (2023-10-14)[2025-09-30]. arXiv: arXiv.2305.15062.

Lu W, Luu R K, Buehler M J. Fine-tuning large language models for domain adaptation: exploration of training strategies, scaling, model merging and synergistic capabilities[J]. npj Computational Materials, 2025, 11: 84.

Ermon S, Finn C, Manning C D, et al. Direct preference optimization: your language model is secretly a reward model[C]//Proceedings of the Advances in Neural Information Processing Systems 36. Neural Information Processing Systems Foundation, Inc. (NeurIPS), 2023: 53728-53741.

王京鲁, 李杰, 冯晓川, 等. 基于人工智能的森林火灾发生预测研究进展[J]. 陆地生态系统与保护学报, 2025, 5(3): 81-89.

Wang J L, Li J, Feng X C, et al. Research progress on the prediction of forest fire occurrence based on artificial intelligence[J]. Terrestrial Ecosystem and Conservation, 2025, 5(3): 81-89.

崔晓辰, 雷一东. 基于深度学习的林业害虫智能化检测方法研究进展[J]. 世界林业研究, 2024, 37(4): 53-57.

Cui X C, Lei Y D. Research progress in deep-learning-based intelligent forest pest detection methods[J]. World Forestry Research, 2024, 37(4): 53-57.

谭晶维, 张怀清, 郭梦蕾, 等. 林草行业大模型构建思路与应用前景[J]. 林业科学, 2025, 61(7): 170-181.

Tan J W, Zhang H Q, Guo M L, et al. Construction ideas and application prospects of large models in forestry and grassand industry[J]. Scientia Silvae Sinicae, 2025, 61(7): 170-181.

Xie Y, Jiang B W, Mallick T, et al. WildfireGPT: tailored large language model for wildfire analysis[PP]. V4. (2025-04-23)[2025-09-30]. arXiv: arXiv.2402.07877.

Du S Q, Tang S J, Wang W X, et al. Tree-GPT: modular large language model expert system for forest remote sensing image understanding and interactive analysis[J]. The International Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2023, 48: 1729-1736.

Tan J W, Zhang H Q, Yang J, et al. ForestryBERT: a pre-trained language model with continual learning adapted to changing forestry text[J]. Knowledge-Based Systems, 2025, 320: 113706.

Sun J Y, Luo Z Z. ForPKG: a framework for constructing forestry policy knowledge graph and application analysis[PP]. V2. (2025-04-29)[2025-09-30]. arXiv: arXiv.2411.11090.

Xu C, Sun Q F, Zheng K, et al. Wizardlm: Empowering large language models to follow complex instructions[C]//International Conference on Learning Representations (ICLR). Vancouver: ICLR, 2024: 43495-43516.

Liu A X, Feng B, Xue B, et al. DeepSeek-v3 technical report[PP]. V2. (2025-02-18)[2025-09-30]. arXiv: arXiv.2412.19437.

Wang Y M, Luo Y. Integration of correlations with standard deviations for determining attribute weights in multiple attribute decision making[J]. Mathematical and Computer Modelling, 2010, 51(1/2): 1-12.

Pearson K. Note on regression and inheritance in the case of two parents[J]. Proceedings of the Royal Society of London, 1895, 58: 240-242.

Wu G, Zhu J. Multi-label classification: do hamming loss and subset accuracy really conflict with each other[C]//Advances in Neural Information Processing Systems. Massachusetts: MIT Press, 2020: 3130-3140.

Lin C Y. ROUGE: a package for automatic evaluation of summaries[C]//Proceedings of the Text Summarization Branches Out. Stroudsburg: ACL, 2004: 74-81.

Team Q. Qwen2 technical report[PP]. V4. (2024-09-10)[2025-09-30]. arXiv: arXiv.2407.10671.

Yang A, Li A F, Yang B S, et al. Qwen3 technical report[PP]. V1. (2025-05-14)[2025-09-30]. arXiv: arXiv.2505.09388.

Zheng Y W, Zhang R C, Zhang J H, et al. LlamaFactory: unified efficient fine-tuning of 100+ language models[C]//Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics (Volume 3: System Demonstrations). Stroudsburg: ACL, 2024: 400-410.

Hu E, Shen Y L, Wallis P, et al. LoRA: low-rank adaptation of large language models[C]//International Conference on Learning Representations (ICLR). Vancouver: ICLR, 2022: 12513-12525.

Chen J L, Xiao S T, Zhang P T, et al. M3-embedding: multi-linguality, multi-functionality, multi-granularity text embeddings through self-knowledge distillation[C]//Proceedings of the Findings of the Association for Computational Linguistics ACL 2024. Stroudsburg: ACL, 2024: 2318-2335.

HURST A, Lerer A, Goucher A P, et al. GPT-4o system card[PP]. V1. (2024-10-25)[2025-09-30]. arXiv: arXiv.2410.21276.

Luo Z Y, Xu C, Zhao P, et al. WizardCoder: empowering code large language models with evol-instruct[C]//International Conference on Learning Representations (ICLR). Vancouver: ICLR, 2024: 3190-3210.

Sharma K, Kumar P, Li Y Q. OG-RAG: ontology-grounded retrieval-augmented generation for large language models[PP]. V1. (2024-12-12)[2025-09-30]. arXiv: arXiv.2412.15235.

Edge D, Trinh H, Cheng N, et al. From local to global: a graph RAG approach to query-focused summarization[PP]. V2. (2025-02-19)[2025-09-30]. arXiv: arXiv.2404.16130.

Bosma M, Chi E, Ichter B, et al. Chain-of-thought prompting elicits reasoning in large language models[C]//Proceedings of the Advances in Neural Information Processing Systems 35. New Orleans: Curran Associates, Inc., 2022: 24824-24837.

Cao Y, Griffiths T, Narasimhan K, et al. Tree of thoughts: deliberate problem solving with large language models[C]//Proceedings of the Advances in Neural Information Processing Systems. New Orleans: Curran Associates, Inc., 2023: 11809-11822.

Liao H R, Hu S H, Zhu Z H, et al. Forest for the trees: overarching prompting evokes high-level reasoning in large language models[C]//Proceedings of the 2025 Conference of the Nations of the Americas Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers). Stroudsburg: ACL, 2025: 1433-1453.

Abdelaziz I, Basu K, Agarwal M, et al. Granite-function calling model: introducing function calling abilities via multi-task learning of granular tasks[C]//Proceedings of the 2024 Conference on Empirical Methods in Natural Language Processing: Industry Track. Stroudsburg: ACL, 2024: 1131-1139.

Zhao W K, Hua J, Wang X J, et al. ToolPlant: tool-based natural language interface for plant simulation models[C]//Proceedings of the 2025 IEEE International Conference on Systems, Man, and Cybernetics (SMC). Piscataway: IEEE Press, 2025: 5673-5678.

Liu W, Huang X, Zeng X, et al. ToolACE: winning the points of LLM function calling[C]//Proceedings of the International Conference on Learning Representations (ICLR). Vancouver: ICLR, 2025: 41359-41381.

陈振升, 罗陶然, 段佳丽, 等. 智慧林业科技前沿与创新发展研究[J]. 西南林业大学学报, 2025, 45(11): 199-208.

Chen Z S, Luo T R, Duan J L, et al. Research on the technological frontiers and innovative development of smart forestry[J]. Journal of Southwest Forestry University, 2025, 45(11): 199-208.

Pretzsch H, Biber P, Schütze G. Forest stand growth dynamics in Central Europe: empirical results and model approaches[J]. Forest Ecology and Management, 2015, 336: 252-264.

Yousefpour R, Hanewinkel M, Jacobsen J B. Modeling adaptive forest management strategies[J]. Forest Policy and Economics, 2012, 24: 16-26.

Pan Y D, Birdsey R A, Fang J Y, et al. A large and persistent carbon sink in the world's forests[J]. Science, 2011, 333(6045): 988-993.

浏览量

下载量

CSCD

文章被引用时，请邮件提醒。

提交

工具集

关联资源

智能的本质与意义

智能时代的“呼吸主权”：从“创世纪计划”观全球科技竞争逻辑

人工智能在基坑工程中的应用综述

RadiFlow：AI Agents重构放射科工作流程

Cerberus：基于深度学习的跨站社交网络机器人检测系统