System Design and Algorithm Optimization of Large Language Model Based on Knowledge Graph in Rumor Detection

The widespread dissemination of misinformation makes finding new methods that combine unstructured text mining and relational data critically important. This work introduces KG-RumorDetect, a framework that combines Knowledge Graphs (KGs) with Large Language Models (LLMs) for improved rumour detection. We employ transformer-based language modelling and sophisticated Knowledge Graph integration, graph neural networks, TransE, DistMult and RotatE knowledge graph embeddings, multi-modal fusion, and end-to-end joint learning. To enhance performance and efficiency further, we apply optimisation techniques such as knowledge distillation, model pruning, quantisation, mixed precision training, dynamic learning rate gradient clipping, and sparse attention with gradual weight adjustment. Extensive experiments confirm that KG-RumorDetect can achieve high detection accuracy while significantly lowering the computational cost, making it a reliable and efficient option for real-time misinformation detection.