Knowledge Reasoning of Large Language Models Integrating Graph-Structured Information for Pest and Disease Control in Tobacco

TL;DR

The paper tackles knowledge-driven pest and disease control in tobacco by integrating a tobacco-specific knowledge graph with a large language model using GraphRAG. It constructs a KG with over $1000$ entities/relations, applies TransE for initial embeddings ($100$-dimensional) and a 2-layer GCN to capture relational structure, and fuses these graph representations with a ChatGLM LLM via LoRA (rank $16$) for efficient fine-tuning. The approach yields superior performance across direct, multi-hop, and comparative reasoning tasks, outperforming baselines such as RAG and KGE+LLM and showing particular strength in complex reasoning scenarios. This graph-augmented reasoning framework provides stronger, domain-specific decision support for agriculture and can be extended to other knowledge-intensive domains like healthcare.

Abstract

This paper proposes a large language model (LLM) approach that integrates graph-structured information for knowledge reasoning in tobacco pest and disease control. Built upon the GraphRAG framework, the proposed method enhances knowledge retrieval and reasoning by explicitly incorporating structured information from a domain-specific knowledge graph. Specifically, LLMs are first leveraged to assist in the construction of a tobacco pest and disease knowledge graph, which organizes key entities such as diseases, symptoms, control methods, and their relationships. Based on this graph, relevant knowledge is retrieved and integrated into the reasoning process to support accurate answer generation. The Transformer architecture is adopted as the core inference model, while a graph neural network (GNN) is employed to learn expressive node representations that capture both local and global relational information within the knowledge graph. A ChatGLM-based model serves as the backbone LLM and is fine-tuned using LoRA to achieve parameter-efficient adaptation. Extensive experimental results demonstrate that the proposed approach consistently outperforms baseline methods across multiple evaluation metrics, significantly improving both the accuracy and depth of reasoning, particularly in complex multi-hop and comparative reasoning scenarios.