One Computer Science MSc thesis to be presented on 9 June

Monday, 9 June there will be a master thesis presentation in Computer Science at Lund University, Faculty of Engineering.

The presentation will take place in E:4130 (Lucas).

Note to potential opponents: Register as an opponent to the presentation of your choice by sending an email to the examiner for that presentation (firstname.lastname@cs.lth.se). Do not forget to specify the presentation you register for! Note that the number of opponents may be limited (often to two), so you might be forced to choose another presentation if you register too late. Registrations are individual, just as the oppositions are! More instructions are found on this page.

15:15-16:00 in E:4130 (Lucas)

Presenters: Ola Berglund, Hampus Mattfolk
Title: Automating Account Linking for Fraud Detection in E-Commerce
Examiner: Jacek Malec
Supervisors: Leonard Papenmeier (LTH), José Lorenzo (technical advisor, Boozt AB), Hanui Ye (business integrator, Boozt AB)

E-commerce platforms grapple with the challenge of users creating multiple accounts to circumvent restrictions that were imposed in response to policy-violating behavior. Current methods, like inflexible 'hard links' (exact attribute matches) or labor-intensive manual 'soft links' (judgment-based associations), are often inadequate. This thesis addresses the critical need for automating account linking to enhance fraud detection capabilities. We develop and evaluate machine learning models, comparing feature-based Random Forests against embedding-based Siamese Neural Networks with BiLSTM encoders. Using internal e-commerce data, models were trained on labeled connections to predict whether different user profiles belong to the same actor. The research investigates strategies for managing imbalanced datasets and optimizing model performance across various precision-recall trade-offs, and demonstrates how these machine learning techniques improve fraud detection efficiency and accuracy by automating account linkage. It also enables analysts to dynamically adjust the precision-recall balance with confidence to suit operational priorities.

Link to popular science summary: To be uploaded