Entity Resolution with Weighted Constraints

Abstract

Constraints ubiquitously exist in many real-life applications for entity resolution (ER). However, it is always challenging to effectively specify and efficiently use constraints when performing ER tasks. In particular, not every constraint is equally effective or robust, and using weights to express the “confidences” on constraints becomes a natural choice. In this paper, we study entity resolution (ER) (i.e., the problem of determining which records in a database refer to the same entities) in the presence of weighted constraints. We propose a unified framework that can interweave positive and negative constraints into the ER process, and investigate how effectively and efficiently weighted constraints can be used for generating ER clustering results. Our experimental study shows that using weighted constraints can lead to improved ER quality and scalability.