Enhancing Knowledge Graph Completion with Retrieval-Augmented Generation Using Large Language Models

Knowledge Graphs (KGs) play a critical role in organizing structured data for applications such as semantic search, recommendation systems, and Artificial Intelligence (AI) analytics. However, the inherent incompleteness of KGs limits their utility and effectiveness. This study introduces an innovative framework for Knowledge Graph Completion (KGC) leveraging the capabilities of Large Language Models (LLMs) through a Retrieval-Augmented Generation (RAG) approach. The proposed methodology treats KG triples as textual prompts to retrieve and integrate relevant information from extensive knowledge bases. By employing entity and relation representations, the LLM is guided to generate contextually accurate and semantically enriched responses. A retrieval reranking strategy further refines predictions by incorporating outputs from a pre-trained KGC model. The framework's performance is evaluated on benchmark datasets using Mean Reciprocal Rank (MRR) and Hits@10 metrics, demonstrating superior precision, scalability, and adaptability compared to existing methods. This research significantly advances the reliability and functionality of KGs, enabling broader applications in AI-driven systems and dynamic data platforms.