Knowledge Graph Interoperability and Reasoning: A Brief Survey of Alignment Techniques and Retrieval-Augmented Generation Approaches

This survey investigates the technical mechanisms for achieving interoperability between heterogeneous Knowledge Graphs (KG) and their application in grounding Large Language Models (LLMs). It details the spectrum of alignment methodologies, moving from fundamental terminological and structural measures such as string edit distances and graph homomorphism to advanced representation learning techniques like translational models and Graph Convolutional Networks (GCN). Furthermore, the document analyzes the convergence of structured data with generative AI, specifically examining how frameworks like Retrieval Augmented Generation (RAG) and neuro-symbolic alignment utilize KGs to enforce factual consistency and mitigate hallucinations in Transformer-based architectures.

View on IEEE Xplore

View on Google Scholar

Recommended citation: A. G. Ravishankar, S. Bansal. "Knowledge Graph Interoperability and Reasoning: A Brief Survey of Alignment Techniques and Retrieval-Augmented Generation Approaches." 2026 International Conference on Semantic Computing (ICSC), pp. 409-415, IEEE, 2026.
Download Paper