Federica Cugnasco
Analysis of a multi-channel anti-fraud platform in banking.
Rel. Alessandro Savino, Nicolò Maunero. Politecnico di Torino, Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering), 2023
|
PDF (Tesi_di_laurea)
- Tesi
Licenza: Creative Commons Attribution Non-commercial No Derivatives. Download (3MB) | Preview |
Abstract: |
Banks are facing fraud attacks every day and in order to minimize their exposure they need to implement the latest anti-fraud technologies. In the banking industry, the costs of fraud incurred cause large damages compared to the lower costs of implementing solutions to prevent attacks. Checking on banking transactions, such as a withdrawal, wire transfer, or deposit, must take place both in real time, at the time the transaction is created, and in batch, which often falls within a day. An anti-fraud architecture is a complex system that correlates many components, each based on specific protocols and algorithms, and makes them able to communicate with each other so that transactions can be analyzed both in real time and at a later time. An anti-fraud check consists of many steps, some among them being the extraction of the historical data of the customers involved in the transaction, verification of their presence on any blacklists or whitelists, and a check on the amount of money. Then all the data will be passed to components that will analyze it and produce a risk code. Based on this code, the transaction will be accepted or rejected. The challenge of an anti-fraud architecture is having to run every possible check, on the user, on accounts, on fraud trends, on background and other checks still, managing to produce a risk index within very few seconds. In fact, the bank not only wants to avoid all fraud but also wants to keep a customer satisfied, performing his transactions without taking up time. Included in this time is latency, which leaves very little time for processing and communication of the other components. The modules that make up the architecture are chosen based on their performance, cost, and the ease with which they can be introduced into the system and communicate with adjacent modules. Some components use only incoming and outgoing REST API calls, while others expect to send data via a Kafka or MQ queue. Once suitable components have been identified, these must be integrated into it and tested for efficiency and reliability. |
---|---|
Relators: | Alessandro Savino, Nicolò Maunero |
Academic year: | 2023/24 |
Publication type: | Electronic |
Number of Pages: | 66 |
Subjects: | |
Corso di laurea: | Corso di laurea magistrale in Ingegneria Informatica (Computer Engineering) |
Classe di laurea: | New organization > Master science > LM-32 - COMPUTER SYSTEMS ENGINEERING |
Aziende collaboratrici: | Accenture SpA |
URI: | http://webthesis.biblio.polito.it/id/eprint/29587 |
Modify record (reserved for operators) |