Crop GraphRAG: Pest and Disease Knowledge Base Q&A System for Sustainable Crop Protection

Hao WuNengfu Xie^*Xiaoli WangJing-Chao FanYonglei LiMeng Zhibo^*

• Chinese Academy of Agricultural Sciences, Beijing, China

Intelligent prevention and control of crop diseases and pests is a critical link in safeguarding food security. However, agricultural practitioners often face fragmented information and low retrieval efficiency when seeking accurate, actionable knowledge, and general-purpose large language models are prone to inaccurate or erroneous answers in specialized domains. To address this problem, we assembled a large-scale corpus of knowledge on crop diseases and pests and, via entity and relation extraction, constructed a multi-relational knowledge graph covering crops, diseases, pests, symptoms, and control measures. By retrieving adjacency subgraphs for the relevant entities together with summary-based passage retrieval, we designed Crop GraphRAG, a new framework that integrates knowledge graphs with retrieval-augmented generation (RAG), thereby enabling local knowledge-base question answering for crop disease and pest management. To evaluate performance, we curated a domain-specific test suite of question–answer pairs and conducted comparative and ablation experiments, which demonstrate advantages in answer accuracy and coverage while effectively suppressing hallucinated content. These results verify the practical utility of the Crop GraphRAG 2 framework for vertical-domain question answering, mitigate the limitations of large language models in specialized agricultural contexts, and provide a pragmatic tool for intelligent QA in the agricultural domain.