Abstract: Objectives: The rapid advancement of artificial intelligence, remote sensing, and IoT-based electronic information technologies has driven multidisciplinary integration in geological disaster monitoring and early warning. Traditional review methods are inadequate for systematically assessing the developmental trends of this field. This work aims to introduce a novel knowledge discovery approach using large language models and structured prompt engineering to comprehensively analyze the knowledge structure, technological paradigms, and evolutionary trends in this domain. Methods: A six-dimensional analytical framework—encompassing disaster types, monitoring technologies, equipment, early warning systems, monitoring data, and algorithmic models—was developed leveraging LLMs. This framework was applied to analyze 242 core publications from 2023 to 2025, extracting key knowledge entities and relationships to construct a comprehensive knowledge graph. Results: The topological analysis reveals a dominant “Landslide-InSAR-LSTM” technological chain, which effectively bridges the gap between discontinuous satellite observations and continuous disaster evolution. Crucially, a divergence in monitoring strategies is identified: research on plastic failures prioritizes long-term surface deformation monitoring, while brittle failures focus on capturing internal rupture signals via microseismic and acoustic emission techniques. Future trends indicate a shift toward “air-space-ground-deep” multi-source integration and hybrid intelligent models that combine physical mechanisms with data-driven algorithms. Conclusions: The proposed analytical model effectively elucidates the knowledge structure and developmental trajectories of this interdisciplinary field, providing a robust scientific foundation for strategic technological roadmapping. Additionally, this approach offers a generalizable methodology for automated knowledge discovery in complex domains.