Fraud Detection Demo

This demo shows how yFiles for HTML can be used for detecting fraud cases in time-dependent data. Fraud affects many companies worldwide causing economic loss and liability issues. Fraud detection relies on the analysis of a huge amount of data-sets and thus, visualizations can be valuable for the quick detection of fraud schemes.

Main Graph Component

Shows the graph according to the current time frame.
Provides additional information for the nodes on right-click.
Highlights fraud rings on hover.
Updates the highlights/selection in the timeline when hovering/selecting non-fraud nodes.
Comes with a special zooming mode: Shift-press zooms out and Shift-release zooms to the current mouse location.

Timeline Component

Shows the number of node creation/removal events with a bar for each point in time.
Contains a time frame rectangle to select which time segment is represented in the main graph by resizing/dragging it or by double-clicking at a specific location.
Provides three detail levels (days/months/years) that are switched by scrolling anywhere in the component.
Updates the highlights/selection in the main graph when hovering/selecting bars.
Offers a -button to automatically move the time frame to the right while updating the main graph.

Detailed View Component

Opens a detailed view of a fraud ring when clicking on fraud ring elements or the according -symbol in the toolbar. Also, hover on a -symbol animates the view port to the corresponding fraud component.
Shows a single graph component that contains fraud rings along with its own timeline.
Updates the layout when clicking the Layout-button in the toolbar.

Fraud Ring

About First-party Bank Fraud

In a first-party bank fraud scenario, fraudsters request legal products from banks, i.e., new accounts, checks, loans or credit cards. For some period of time, they behave like normal customers and pay their debts. However, suddenly they disappear with the money leaving no trace behind, since they have used fake identities or contact information.

In our graph, each person is associated with an address, a phone number, a bank branch and a series of bank products (including possible payments for these products). Each node is visualized in a different manner based on its type and is associated with dates representing the date when an event started and/or finished. For example, when a loan was requested and/or was paid back.

A typical fraud-scenario involves two or more persons that share the same fake personal information such as address or phone number and apply for several bank products.

What to look for

Fraud rings, i.e., persons that form cycles and share contact information. In the demo, they are visualized in red color.

Loading... This may take a while.