Link analysis
Link analysis is a data science technique in which the different nodes of a network are analyzed to identify similarities and relationships. These links are often portrayed visually.
Through link analysis, it’s possible to understand how different accounts on a platform are related to one another based on the information or account details they share, as well as the actions each account takes. It’s also possible to look for patterns and anomalies in a given network, which may be indicative of fraud.
Frequently asked questions
How is link analysis used to fight fraud?
If link analysis uncovers multiple accounts on your platform or database sharing suspicious account details, that may be indicative of fraud. Your investigations team can flag this account cluster for further review or block them outright. They can also prevent new accounts from being opened using any of the properties found in the flagged accounts.
Similarly, if you discover a known fraudster in your database, you can use link analysis to uncover other accounts on your platform sharing the same properties, such as their IP address or phone number.
What kind of connections between accounts may be indicative of fraud?
There are many different signals that could be considered suspicious through the lens of link analysis. Some of the most important signals include shared account details such as:
- Physical address
- IP address
- Device fingerprint
- Browser fingerprint
- Payment details
It’s important to note, however, that there may be legitimate reasons for multiple accounts to share certain details. Members of a family living at the same address may all share the same IP address, device fingerprint, etc. if they each open an account on the same website, for example. To differentiate between legitimate and suspicious links, businesses must look more broadly at all properties within a cluster and assess from there.