Graph+AI for Fin. Services

Graph+AI for Fin. Services
September, 2020
1

. | GRAPHAIWORLD.COM |
Today’s Presenter
2
Abhishek Mehta
Director of Sales Engineer
● McKinsey, Bloomberg, Cisco & Dabizmo (NLP Startup) Founder
● 15+ years designing and implementing complex analytics
solutions for Fortune 100 companies
● Patents in NLP spanning Conceptunary Ontology Design,
Language Pattern Recognition, and Conversion

Behind the Challenges
3
© https://www.postgresqltutorial.com/postgresql-data-types/

Lies Our Struggle To Survive, Thrive, Innovate & Win
4
© https://www.bluesheep.com/blog/five-key-b2b-marketing-strategies

Lies Our Struggle To Survive, Thrive, Innovate & Win
5
Financial
Institutions are the
Number 1
adopters of
technological
Innovation, and
graph is no
different.

6 of the Top
10 Global
Banks Use
TigerGraph Merchant Analytics:
Transaction
sequencing to
detect geolocation
proximity.
Credit Card Fraud
Is applicant
connected to
potential
fraudsters?
Trade
Surveillance: Are
employees
following the
rules?
Impact Analysis
Communities or
Clusters
impacted by
the fraud rings
Credit Scoring
Real-time credit
scoring to help
recommend
offers best suited
to customer
profiles?
6
Wealth
Management:
What Accounts,
HNI to target for
stocks or life
change events.

© 2020 TigerGraph. All Rights
Reserved
Innovation Stunting due to Data Issues
Deep Wide Operational
Analyze relationships
deeper into the data
to find hidden fraud
patterns
Connect all datasets to
uncover undisclosed
relationships
Process transactions in
real-time to flag
possible fraud
At Scale
7

Reserved
Data Engineering View
8
Data Quality
Data Connectivity
(1) Diverse Data Sources
(2) Not Interlinked
(3) Privacy & Security on
Connectivity
(4) Hence,No Ad-Hoc
Querying
(5) Poor Performance on
Speed And Scale
(1) Diverse Data Formats
(2) Millions of Entities
(3) Thousands of
Attributes
(4) Data Consistency
Connected Data
Perform at Scale
Perform at Speed
Cover Future Needs
Download the solution brief at - https://info.tigergraph.com/MachineLearning
Choosing Graph

© 2020 TigerGraph. All Rights Reserved
Core Technology Groups
Machine Learning, Data Science Groups,
Enterprise Architecture, Infrastructure
3
Modelling Group
Statisticians, Domain Experts, Play/Rule books,
Experts in Credit, Fraud Behaviour
2
Operational Groups
Fraud Analyst, Compliance , Regulations,
Forecasting, Reporting, Visualizatoins
1
Enterprise Architecture View
9

. | GRAPHAIWORLD.COM | 10
Monica Rogati, an equity partner at Data Collective
6 of the Top
10 Global
Banks Use
TigerGraph

Reserved
Relational Database Key-Value Database Graph Database
Customer
XXXXXX
Product
XXXXXXXX
Supplier
XXXXXXXX
Location
XXXXXXXX
Order
XXXXXXXXX
Product
Customer
Supplier
Location
KEY VALUE
XXXXX
Order
Customer
Prod
uct
Supplier
• Rigid schema
• High performance for transactions
• Poor performance for deep analytics
• Highly fluid schema/no schema
• High performance for simple transactions
• Poor performance deep analytics
• Flexible schema
• High performance for complex
transactions
• High performance for deep analytics
Location 1 = Delivery Location
Location 2 = Warehouse
Location 2
Product
Payment
PURCHASE
D
RESIDES
SHIPSTO
PURCHASE
D
SHIPS FROM
A
C
C
EPTED
MAKES
Graph As A Solution
XXXXX
XXXXX
XXXXX Location 1
N
O
TIFIES
11

Semantics of the Use Case
12
Receive Transaction
Does TransactionUses
Device
UsesPI
User
Device
Transaction
Payment Instrument

Deep Link Multi-Hop Analytics
Email 1
Has
Account 1
A
uthenticated_by
Phone_number 1
User 1
Authenticated_by
Has
Device 1 (Apple iPhone 6)
Linked_to
Used_with
Credit Card
From
Send_payment
Payment 1
Used_for
Receive_paym
ent
Account 2
User 2
Sets_Up
Sets_Up
From
Bank
Device 101
(Samsung Galaxy S3)
User 101
Account 101
Stolen Credit Card
Used_with
Phone_number 101
Used_with
Sets_Up
Has
Authenticated_by
Phone_number 2
Hop 1
Hop 2
Hop3
Hop4
Hop5
Used_for
Payment 101
Send_paym
ent
Hop6
13
Learn more by reading this article: How the World’s Largest Banks Use Graph Analytics to Fight Fraud
Sign up FREE for TigerGraph Cloud to use the starter kit for fraud detection (payments)

Demo
14

7 Key Data Science Capabilities Powered By a Native Parallel Graph

Reserved
HTAP: Hybrid Transaction/Analytical Processing
Zero Latency & Zero Friction Between Transactions and Analytics
+ Real-time read & write
+ Real-time, compute-intensive,
multi-dimensional analysis
+ Real-time aggregation
OLTP and OLAP Together in Graph
Some Graph Databases
+ ACID
+ Concurrency
All Graph Databases
o Multi-dimensional data
o Real-time read
16
OLTP - Transactional
● Real-time read and write
● ACID properties
(guarantee that transaction is
correct)
● Concurrency
(many transactions at the same time)
OLAP - Analytical
● Multi-dimensional Analysis
● Compute-intensive
● Data-intensive
● Aggregation

© 2020. ALL RIGHTS RESERVED. | TIGERGRAPH.COM | CONFIDENTIAL INFORMATION |
Detecting Fraud Rings with TigerGraph
17
Business Challenge
A leading U.S. bank wanted a better way to detect and remove
fraudsters from their credit-card network. Prototypes had shown
that a combination of advanced graph algorithms gave significant
gains – big-data tools and other graph technologies either couldn’t
scale to the full customer base or gave inconsistent results.
Solution
• Implementing PageRank and Louvain [fraud] community
detection in an MPP native-parallel database.
• Leveraging deep analytics to find hidden connections across
20TB+ of data.
Business Benefits
• Able to expose fraud rings, shut down connected cards, and
combat fraudulent activity on a massive scale –35% uplift
and $50M incremental fraud avoidance. >$1.5 million through
cost savings on false positives, infrastructure and TCO
Tier 1 U.S. Bank
10TB
Card applications
data
6 weeks
PoC elapsed time
3 months
Time to build and fully deploy
platform to production
+$50M
1st
year ROI with 35%
uplift in fraud detection
CLV Impact > $200M

© 2020. ALL RIGHTS RESERVED. | TIGERGRAPH.COM | CONFIDENTIAL INFORMATION |
Preventing Fraudulent Loans with
TigerGraph
18
Business Challenge
A leading U.S. bank needed to search across 20TB of data for
possible connections between credit card applications known to
be fraudulent and applications of unknown status - relational
databases and other graph providers were not up to the task, as
they were unable to deliver the speed and scale required.
Solution
• Pairing graph technology with machine learning to identify
fraudulent activity at scale and intervene in real-time.
• Leveraging deep analytics to find hidden connections
across 20TB+ of data.
Business Benefits
• Able to score and prevent fraudulent loan applications on a
massive scale – minimum 30% uplift and $15M annual
incremental fraud avoidance. $1.5M through cost savings
on false positives.
Tier 1 U.S. Bank
5TB
Card applications
data
6 weeks
PoC elapsed time
3 months
Time to build and fully deploy
platform to production
$16.5M
1st
year ROI with 30%
uplift in fraud detection
CLV Impact > $100M

Intuit Implements Real-Time Fraud Investigation
19
Business Challenge
Intuit realized their homegrown link analysis solution for fraud detection
needed an upgrade. Risk investigators were switching between
multiple tools causing a manual, time consuming process.
Solution
• Payments with potential fraudulent activity are flagged in
real-time
• Over 50 investigators explore each suspected transaction using
GraphStudio to understand linkage to customer accounts and
previous transactions with fraudulent activity
Business Benefits
Fraud Detection solution with TigerGraph has improved the
productivity and efficiency of the risk investigators, allowing them to
identify and investigate fraud in real-time connecting information
across the organization.
Learn about TigerGraph’s customers: https://www.tigergraph.com/customers/

OpenCorporates Upgrades Performance and
Functionality
20
Business Challenge
OpenCorporates, world’s largest open database of corporate information
experienced challenges in terms of scalability, lack of support for simple
queries and speed as it ramped up to production using a first-generation
graph technology
Solution
• Support queries of up to five degrees of separation to help uncover
relationships between entities and see which relationships are active vs.
dead
• Unlock insight into how relationships and networks have changed and
evolved over time (temporal graph search)
Business Benefits
OpenCorporates is scaling up its database with 170 million corporate
entities to provide users with deeper analysis of the information, helping
them uncover instances of criminal or anti-social activity - such as
corruption, money laundering, and organized crime“As our work continues and our data grows, we had challenges scaling our
data to meet our business needs. TigerGraph’s excellent scalability and
performance enables us to achieve things we previously could not do, and to
better support ongoing investigative work in the process.”
Additional details in the press release -https://tinyurl.com/y36skysr
20

IceKredit Builds A Customer 360 Graph For
Credit Rating and Risk Assessment
21
Business Challenge
With rapid growth in size and complexity of the interconnected global
financial markets making it difficult for banks to process loan applications
for home, automobile, etc.
Solution
• Leverage Machine Learning and AI for custom advanced models and
analytics to build comprehensive credit views for applicant
• Quantify applicant’s fraud probability and compares it with actual business
activity
• Find undisclosed relationships and connections within data; assign and
update risk ratings in real time
Business Benefits
IceKredit is empowering lenders by reducing their fraud risks with more
accurate, detailed credit ratings for applicants that are not tracked by
traditional credit bureaus.
Additional details in the TechTarget article - https://tinyurl.com/y696vkqf
21

Pagantis Delivers Faster Consumer Finance
Services with TigerGraph on AWS
Business Challenge
Pagantis must assess credit worthiness and fraud risk in real-time for
customers to allow them to pay for their purchase in monthly
installments. Risk assessment with relational databases was taking too
long, delaying the time for loan approvals.
Solution
• Real-time calculation of customer’s credit rating using their
current activities as well as all available historical data
• A scalable, high-performance system to deliver insights into
complex relationship-based workflows for credit scoring, fraud
detection, recommendation engines and risk analysis
Business Benefits
Pagantis can now offer a faster and seamless consumer finance
solution for the eCommerce merchants throughout Italy, France and
Spain.
Press Release - https://info.tigergraph.com/pagantis-tigergraph
22

TigerGraph - Product Overview & Differentiators
1. A Native Parallel and Distributed Graph Database - Graph 3.0, MPP architecture
2. Scalability - Scale Up and Scale Out
3. Performance - Analytical and ML performance combined with Up/Insert Performance
4. Enterprise Features - Delivery modes, High Availability, Security, API Integration
5. GSQL - SQL-like, easy-to-learn, high performance query language that is Turing complete.
6. Deep Link Analytics - traverse graph in parallel, filter and do graph computations, 12+ hops
in production
7. Graph Algorithms Library - open source on github, flexibility & rapid time to value
8. MultiGraph service for multi tenancy with RBAC permissioning
9. GraphStudio - Comprehensive visual SDK; from graph design to deployment
10. Advanced data compression technology - approximately 50%, better TCO
11. Machine learning feature generation - generate deep-link (multi-hop) graph features in
real-time for massive datasets for feeding ML solutions
12. ACID Compliant - can do both OLTP + OLAP on single platform, cost effective

Graph+AI for Fin. Services

Empfohlen

Empfohlen

Weitere ähnliche Inhalte

Was ist angesagt?

Was ist angesagt? (20)

Ähnlich wie Graph+AI for Fin. Services

Ähnlich wie Graph+AI for Fin. Services (20)

Mehr von TigerGraph

Mehr von TigerGraph (16)

Kürzlich hochgeladen

Kürzlich hochgeladen (20)

Graph+AI for Fin. Services