FairEM360: A Suite for Responsible Entity Matching

TL;DR

FairEM360 tackles the risk of bias in entity matching by introducing a dedicated fairness auditing suite. It operationalizes both single and pairwise EM fairness with a suite of five EM-specific group fairness definitions and multiple disparity measures, while enabling explainability and ensemble-based resolutions across ten integrated matchers. The framework uses a three-layer architecture (data, logic, presentation) and containerized matchers to support scalable, interactive auditing and remediation. The demonstration highlights practical workflows on real datasets, illustrating how practitioners can identify unfair groups, reason about causes, and select fairer matching strategies.

Abstract

Entity matching is one the earliest tasks that occur in the big data pipeline and is alarmingly exposed to unintentional biases that affect the quality of data. Identifying and mitigating the biases that exist in the data or are introduced by the matcher at this stage can contribute to promoting fairness in downstream tasks. This demonstration showcases FairEM360, a framework for 1) auditing the output of entity matchers across a wide range of fairness measures and paradigms, 2) providing potential explanations for the underlying reasons for unfairness, and 3) providing resolutions for the unfairness issues through an exploratory process with human-in-the-loop feedback, utilizing an ensemble of matchers. We aspire for FairEM360 to contribute to the prioritization of fairness as a key consideration in the evaluation of EM pipelines.

Figures (7)

• Figure 1: FairEM360 Architecture
• Figure 2: data import step.
• Figure 3: matcher selection step.
• Figure 4: audit step. evaluation criteria (left) and audit results (right)
• Figure 5: unfairness explanations for the observed in the cn group w.r.t. TPRP.