AI-Driven Entity Resolution: Enhancing Customer Data Matching with Explainable Graph Learning

Abstract

Accurate entity resolution is essential for maintaining high-quality customer data in enterprise systems. This study presents an explainable AI-driven approach to entity matching using Graph Neural Networks (GNNs) for structured and relational customer data. We introduce a novel Explainable Entity Matching (xEM) framework that improves transparency in data linkage by leveraging node embeddings and probabilistic matching techniques. Our approach is evaluated against existing entity resolution methods, including heuristic-based models and deep learning architectures, across real-world and synthetic datasets. Experimental results demonstrate that xEM enhances accuracy and interpretability, reducing false positives in transitive linking while maintaining scalability for large datasets. This work provides insights into optimizing AI-driven data management strategies for enterprise applications.