Resolvi: A Reference Architecture for Extensible, Scalable and Interoperable Entity Resolution

Abstract

Context: Entity resolution (ER) plays a pivotal role in data management by determining whether multiple records correspond to the same real-world entity. Because of its critical importance across domains such as healthcare, finance, and machine learning and its long research history designing and implementing ER systems remains challenging in practice due to the wide array of methodologies and tools available. This diversity results in a paradox of choice for practitioners, which is further compounded by the various ER variants (record linkage, entity alignment, merge/purge, a.s.o). Objective: This paper introduces Resolvi, a reference architecture for facilitating the design of ER systems. The goal is to facilitate creating extensible, interoperable and scalable ER systems and to reduce architectural decision-making duration. Methods: Software design techniques such as the 4+1 view model or visual communication tools such as UML are used to present the reference architecture in a structured way. Source code analysis and literature review are used to derive the main elements of the reference architecture. Results: This paper identifies generic requirements and architectural qualities of ER systems. It provides design guidelines, patterns, and recommendations for creating extensible, scalable, and interoperable ER systems. Furthermore, it highlights implementation best practices and deployment strategies based on insights from existing systems. Conclusion: The proposed reference architecture offers a foundational blueprint for researchers and practitioners in developing extensible, interoperable, and scalable ER systems. Resolvi provides clear abstractions and design recommendations which simplify architecture decision making, whether designing new ER systems or improving existing designs.