Electre Tri-Machine Learning Approach to the Record Linkage Problem

TL;DR

This paper introduces a novel application of the Electre Tri-Machine Learning method to the record linkage problem, demonstrating high accuracy in identifying matches and nonmatches.

Contribution

It adapts the Electre Tri method, typically used for ordinal classification, to improve record linkage accuracy, which is a new approach in this context.

Findings

Over 99% accuracy in match and nonmatch identification

Electre Tri method effectively applied to record linkage

Preliminary results show promising potential

Abstract

In this short paper, the Electre Tri-Machine Learning Method, generally used to solve ordinal classification problems, is proposed for solving the Record Linkage problem. Preliminary experimental results show that, using the Electre Tri method, high accuracy can be achieved and more than 99% of the matches and nonmatches were correctly identified by the procedure.