Using graph
technologies to fight
fraud.
SAS founded in 2013 in Paris | http://linkurio.us | @linkurious
French startup specialized in graph-visualization.
CTO
Web-scale archiving
Université de
Technologie de
Compiègne
CMO
>5 years in consulting
Sciences Po + Ecole
de Guerre
Economique
Jean
Villedieu
Sébastien
Heymann
David
Rapin
CEO
Created Gephi
Phd in CS and
complex systems
from UPMC
What is a graph?
Father Of
Father Of
Siblings
A graph is a set of nodes and relationships.
This is a node
This is a
relationship
Father Of
Father Of
Siblings
Why graphs are important?
Rise of complex and connected data.
Increase in
volume
New processes, more
transactions, more social
devices, more devices, etc.
Increase in
connectedness
Customers, products,
processes and devices
interact with each other.
Relational DBS are not
good at relationships.
● Cannot model data and relationships without
complexity;
● Performance degrades with number and levels
of relationships, and database size;
● Adding new types of data and relationships
requires schema redesign.
Graph DBs unlock
connected data.
● Store graphs of billions of nodes and edges;
● Query the graph to find interesting patterns;
● Popularity of graph databases has increased
500% in the last 2 years;
● Recommendation, fraud detection, network
management, cybersecurity, health...
Why does it work well for
fraud?
Layer 1 Layer 2 Layer 3 Layer 4 Layer 5
Endpoint-
centric
Navigation-
centric
Analysis of users
and their
endpoints
Account-
centric
Cross-
channels
Entity link
analysis
Analysis of
navigation
behavior and
suspect patterns
Analysis of
anomaly
behavior on a
per-channel
basis
Analysis of
anomaly
behavior
correlated on a
cross-channel
basis
Analysis of
relationships to
detect organized
collusion
Network analysis offers a unique opportunity to identify the sophisticated fraudsters who work under the radar
but in a coordinated fashion.
AML.
Normal customer #1
Normal customer #2
In business with
Suspicious individual Known criminal
In business with
In business with
Are my customers in contact
with criminals?
Conflict of interest.
Are my vendors linked to my
employees?
New vendor
Address
Employee
Registered address
Phone number
Registered phone Registered phone
Registered address
Synthetic identity.
Customer #1
Address
Customer #2
Phone number
Bank loan Bank loan
Am I loaning money to
people who don’t exist?
Swiss Leaks.
Finding the real owners of
$100 billion.
Real owner Pseudo owner Offshore entity Swiss bank account
Controls ControlsMarried
Stolen credit cards.
Customer #1
Merchant #3
Customer #2
Merchant #2
Merchant #1 Merchant #4
Who is stealing credit cards?
Normal TX
Contested TX Contested TX
Normal TX
Contested TX Contested TX
First enterprise-ready graph visualization platform.
Improve your fraud detection system.
Detect new
fraud cases.
Graph databases like Neo4j
provides the ability to identify
fraud patterns at scale.
Faster fraud
investigations.
Linkurious provides a simple
interface to investigate
suspicious patterns.
How it works.
Legacy DB
DB
On-premise server Web-browser
Pricing.
Starter Enterprise Toolkit
Search and explore your
data
Enterprise-ready security
and collaboration
Build or improve an
existing web application
Desktop Server Code
Single user Mature organization Mature organization
€990/user/year €1,990/user/year €30,000/product/year
Conclusion.
Question?
Want to know more?
● Insider trading: https://linkurio.us/using-graphs-to-uncover-insider-trading-schemes/
● Conflict of interest: https://linkurio.us/fraud-detection-identifying-conflicts-of-interest-with-graphs/
● Synthetic identify: https://linkurio.us/how-to-detect-bank-loan-fraud-with-graphs-part-1/ and
https://linkurio.us/how-to-detect-bank-loan-fraud-with-graphs-part-2/
● Ecommerce fraud: https://linkurio.us/reshipping-scams-and-network-visualization/
● Swiss Leaks: https://linkurio.us/how-the-icij-used-linkurious-to-reveal-the-secrets-hidden-in-the-
swiss-leaks-data/
● Fraud detection in retail: https://linkurio.us/fraud-detection-in-retail/
● Credit card fraud: https://linkurio.us/stolen-credit-cards-and-fraud-detection-with-neo4j/

Using graph technologies to fight fraud

  • 1.
    Using graph technologies tofight fraud. SAS founded in 2013 in Paris | http://linkurio.us | @linkurious
  • 2.
    French startup specializedin graph-visualization. CTO Web-scale archiving Université de Technologie de Compiègne CMO >5 years in consulting Sciences Po + Ecole de Guerre Economique Jean Villedieu Sébastien Heymann David Rapin CEO Created Gephi Phd in CS and complex systems from UPMC
  • 3.
    What is agraph? Father Of Father Of Siblings
  • 4.
    A graph isa set of nodes and relationships. This is a node This is a relationship Father Of Father Of Siblings
  • 5.
    Why graphs areimportant?
  • 6.
    Rise of complexand connected data. Increase in volume New processes, more transactions, more social devices, more devices, etc. Increase in connectedness Customers, products, processes and devices interact with each other.
  • 7.
    Relational DBS arenot good at relationships. ● Cannot model data and relationships without complexity; ● Performance degrades with number and levels of relationships, and database size; ● Adding new types of data and relationships requires schema redesign.
  • 8.
    Graph DBs unlock connecteddata. ● Store graphs of billions of nodes and edges; ● Query the graph to find interesting patterns; ● Popularity of graph databases has increased 500% in the last 2 years; ● Recommendation, fraud detection, network management, cybersecurity, health...
  • 9.
    Why does itwork well for fraud?
  • 10.
    Layer 1 Layer2 Layer 3 Layer 4 Layer 5 Endpoint- centric Navigation- centric Analysis of users and their endpoints Account- centric Cross- channels Entity link analysis Analysis of navigation behavior and suspect patterns Analysis of anomaly behavior on a per-channel basis Analysis of anomaly behavior correlated on a cross-channel basis Analysis of relationships to detect organized collusion Network analysis offers a unique opportunity to identify the sophisticated fraudsters who work under the radar but in a coordinated fashion.
  • 11.
    AML. Normal customer #1 Normalcustomer #2 In business with Suspicious individual Known criminal In business with In business with Are my customers in contact with criminals?
  • 12.
    Conflict of interest. Aremy vendors linked to my employees? New vendor Address Employee Registered address Phone number Registered phone Registered phone Registered address
  • 13.
    Synthetic identity. Customer #1 Address Customer#2 Phone number Bank loan Bank loan Am I loaning money to people who don’t exist?
  • 14.
    Swiss Leaks. Finding thereal owners of $100 billion. Real owner Pseudo owner Offshore entity Swiss bank account Controls ControlsMarried
  • 15.
    Stolen credit cards. Customer#1 Merchant #3 Customer #2 Merchant #2 Merchant #1 Merchant #4 Who is stealing credit cards? Normal TX Contested TX Contested TX Normal TX Contested TX Contested TX
  • 16.
    First enterprise-ready graphvisualization platform.
  • 17.
    Improve your frauddetection system. Detect new fraud cases. Graph databases like Neo4j provides the ability to identify fraud patterns at scale. Faster fraud investigations. Linkurious provides a simple interface to investigate suspicious patterns.
  • 18.
    How it works. LegacyDB DB On-premise server Web-browser
  • 19.
    Pricing. Starter Enterprise Toolkit Searchand explore your data Enterprise-ready security and collaboration Build or improve an existing web application Desktop Server Code Single user Mature organization Mature organization €990/user/year €1,990/user/year €30,000/product/year
  • 20.
  • 21.
    Want to knowmore? ● Insider trading: https://linkurio.us/using-graphs-to-uncover-insider-trading-schemes/ ● Conflict of interest: https://linkurio.us/fraud-detection-identifying-conflicts-of-interest-with-graphs/ ● Synthetic identify: https://linkurio.us/how-to-detect-bank-loan-fraud-with-graphs-part-1/ and https://linkurio.us/how-to-detect-bank-loan-fraud-with-graphs-part-2/ ● Ecommerce fraud: https://linkurio.us/reshipping-scams-and-network-visualization/ ● Swiss Leaks: https://linkurio.us/how-the-icij-used-linkurious-to-reveal-the-secrets-hidden-in-the- swiss-leaks-data/ ● Fraud detection in retail: https://linkurio.us/fraud-detection-in-retail/ ● Credit card fraud: https://linkurio.us/stolen-credit-cards-and-fraud-detection-with-neo4j/