[PDF]

Fuzzy Similarity Matching of different data points for various personas with a real industry use case.


Tobias Gadsby

14/05/2026

Supervised by Federico Liberatore; Moderated by Oktay Karakus

I currently have a continuing role from my placement at Lloyds Banking Group, working in Economic Crime Prevention. Specifically I work in List Screening, which involves developing a system for fuzzy matching of customers against various Economic Crime Controls. My proposal is to compare the efficacy of fuzzy matching scenarios on large customer datasets (80M+). I will explore a variety of fuzzy matching techniques, including various lexical similarity algorithms, geographical matching, semantic similarity and more. I will compare these on a variety of criteria, including efficiency, accuracy, precision and more. I will also discover and develop new techniques, as well as a test suite to accurately compare their efficacy to established industry standards.


Initial Plan (02/02/2026) [Zip Archive]

Final Report (14/05/2026) [Zip Archive]

Publication Form