The Interaction of Data, Data Structures, and Software in Entity Resolution Systems

摘要

This article discusses entity identity information management (EIIM) as implemented in entity resolution (ER) systems. In particular, it focuses on the interaction among the data and software components in different types of ER system architectures. The ability to create and maintain persistent entity identity structures (EIS) and identifiers over time is a critical process for master data management (MDM) and entity-based information exchange systems. The article describes two EIIM models, attribute-based EIIM and record-based EIIM, the design trade-offs associated with each model, and how the choice of an EIIM model can both limit and extend an ER system's resolution capabilities. The article also shows how both EIIM models have been implemented in the open-source ER system OYSTER, and experimental results showing the difference.