Edge-Native Knowledge Graph and RAG Integration for Advanced Intellectual Property Landscape Mapping

Abstract

The exponential growth of global intellectual property (IP) data has introduced significant challenges in extracting timely, accurate, and context-aware insights for strategic decision-making. Traditional centralized analytics systems are increasingly inadequate due to latency constraints, limited scalability, and poor handling of complex semantic relationships embedded within patent documents and citation networks. This study proposes an edge-native architecture that integrates knowledge graphs (KGs) with retrieval-augmented generation (RAG) models to enable advanced IP landscape mapping. The framework leverages distributed edge computing to process data closer to the source, thereby reducing latency and enhancing real-time responsiveness. Knowledge graphs are employed to represent entities such as patents, inventors, and organizations, along with their interrelationships, enabling structured semantic reasoning. The RAG component enhances this capability by dynamically retrieving relevant contextual information and generating coherent, knowledge-informed outputs for analytical tasks. The proposed system is designed to support scalable, low-latency IP intelligence while maintaining high retrieval accuracy and contextual relevance. Experimental evaluation demonstrates improved performance in terms of response time, information retrieval precision, and system throughput compared to conventional centralized and graph-only approaches. The integration of edge computing with KG–RAG pipelines provides a robust and flexible solution for modern IP analytics, offering significant benefits for research institutions, patent offices, and innovation-driven enterprises. This work contributes a novel architectural paradigm that bridges distributed computing and semantic AI for next-generation intellectual property intelligence systems.