Probabilistic Record Linkage and Deduplication after Indexing, Blocking, and Filtering

TL;DR

This paper reviews probabilistic record linkage and deduplication, emphasizing the impact of indexing, blocking, and filtering, and introduces a new model to better account for these preprocessing steps within the Fellegi-Sunter framework.

Contribution

It proposes a novel probabilistic model that explicitly incorporates indexing and filtering steps into the record linkage process, addressing a gap in existing methods.

Findings

Analyzes the effects of indexing and blocking on linkage accuracy.

Introduces a new model that improves inference after filtering.

Provides insights into optimizing preprocessing for record linkage.

Abstract

Probabilistic record linkage, the task of merging two or more databases in the absence of a unique identifier, is a perennial and challenging problem. It is closely related to the problem of deduplicating a single database, which can be cast as linking a single database against itself. In both cases the number of possible links grows rapidly in the size of the databases under consideration, and in most applications it is necessary to first reduce the number of record pairs that will be compared. Spurred by practical considerations, a range of methods have been developed for this task. These methods go under a variety of names, including indexing and blocking, and have seen significant development. However, methods for inferring linkage structure that account for indexing, blocking, and additional filtering steps have not seen commensurate development. In this paper we review the implications of indexing, blocking and filtering within the popular Fellegi-Sunter framework, and propose a new model to account for particular forms of indexing and filtering.