Large Language Models for Test Case Evolution and Repair
GraphTalk Barcelona - Keynote
1. Graph Talk Barcelona
#1 Database for Connected Data
Dirk Möller
Director Sales CEMEA
dirk@neo4j.com
6/4/19
2. Neo4j GraphTalks
Network & Application Management
• Einführung in Graphdatenbanken und Neo4j (9.30-10.00)
Bruno Ungermann
• Neue Herangehensweisen für Network und Application Mgt mit Graphen (10.00-11.00)
Stefan Kolmar
• Wie werden Graphdatenbank-Projekte mit Neo4j zum Erfolg? (11.00-11.30)
Stefan Kolmar
• Q&A
3. Agenda!
• Impact of Graphs
• State of the Graph
• Three waves
• What‘s enabling all of this?
• AI and Graph
6. 2.6 TB
11.5 million documents
Emails, Scanned Documents,
Bank Statements etc… Person
B
Bank US
Account
123
Person
A
Acme
Inc
Bank
Bahama
s
Address
X
HAS_ACCOUNT
REGISTERED
IS_OFFICER_OF
WITH
NODE
RELATIONSHIP
7. 2.6 TB
11.5 million documents
Emails, Scanned Documents,
Bank Statements etc…
21. Category Defining Use Cases
airbnb
Fraud
Detection
Real-Time
Recommendations
Network & IT
Operations
Master Data
Management
Knowledge
Graph
Identity & Access
Management
22. 10M+
Downloads
3M+ from Neo4j Distribution
7M+ from Docker
Events
400+
Approximate Number of
Neo4j Events per Year
50k+
Meetups
Number of Meetup
Members Globally
Largest pool of graph technologists
50k+
Trained/certified Neo4j
professionals
Trained Developers
25. >50%of enterprises are using
graph databases
As of today
Source: Forrester Vendor Landscape: !
Graph Databases, October 6, 2017!
26. "Neo4j continues to
dominate the graph
database market.”
“69% of enterprises
have, or are planning to
implement graphs over
next 12 months”
October, 2017
“The most widely stated
reason in the survey for
selecting Neo4j was
to drive innovation”
February, 2018
Critical Capabilities for
DBMSA
“In fact, the rapid rise of
Neo4j and other graph
technologies may signal that
data connectedness is
indeed a separate
paradigm from the model
consolidation happening
across the rest of the
NoSQL landscape.”
March, 2018
Graph is a Unique Paradigm!
35. Retail
7 of top 10
Finance
20 of top 25
7 of top 10
Software
Hospitality
3 of top 5
Telco
4 of top 5
Airlines
3 of top 5
Logistics
3 of top 5
76%
FORTUNE 100
have adopted or
piloted Neo4j
37. 37
• Record “Cyber Monday” sales
• About 35M daily transactions
• Each transaction is 3-22 hops
• Queries executed in 4ms or less
• Replaced IBM Websphere commerce
• 300M pricing operations per day
• 10x transaction throughput on half the
hardware compared to Oracle
• Replaced Oracle database
• Large postal service with over 500k
employees
• Neo4j routes 7M+ packages daily at peak,
with peaks of 5,000+ routing operations per
second.
Handling Large Graph Work Loads for Enterprises
Real-time promotion
recommendations
Marriott’s Real-time
Pricing Engine
Handling Package
Routing in Real-Time
40. Data Network Effect
“A product, generally powered by machine learning, becomes smarter
as it gets more data from your users. The more users use your product,
the more data they contribute; the more data they contribute, the
smarter your product becomes.”
— Matt Turck
58. “Increasingly we're learning that you can make
better predictions about people by getting
all the information from their friends and
their friends’ friends than you can from the
information you have about the person
themselves”
— Dr. James Fowler
Relationships Are Often the Strongest
Predictors of Behavior
63. $
Better Decisions
Machine Learning Pipeline
• Engineered features
when you know what
you’re looking for
• Feature extraction
and selection using
graph algorithms
• Graph embeddings to
feed into DL
Graphs add highly predictive
features to models; adding
accuracy without altering current
workflows
Graphs can also infer
relationships and add data
where sparse
65. 12
talks on
Graph-Enhanced AI & ML
recorded at
GraphConnect
+8
talks on
Graph-Enhanced AI & ML
during the
Spring GraphTour
66. Four Pillars of Graph-Enhanced AI
1. Knowledge
Graphs
Context for Decisions
2. Connected
Feature Extraction
Context for Credibility
4. AI Explainability3. Graph
Accelerated AI
Context for Efficiency
Context for Accuracy
72. • Transaction Fraud
• Anti-money laundering (AML)
• Claims Fraud
• Credit Fraud
• Compliance and investigation
72
Improve the Predictive Power of ML in Fighting
Financial Crimes
Machine Learning Pipeline
Data
Machine Learning can help uncover &
learn common traits so we can build
more predictive models
Unfortunately many machine
learning methods rely on flat data
structures and tables
73. Engineering connected features improves Machine Learning by
calculating relationship metrics when you know what’s predictive
For example, adding
how many fraudsters are
in someone’s network
is faster and simpler
using connections
Combat Financial Crimes using Connected Features
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
BANK!
ACCOUNT!
SSN/ ID NUMBER!
UNSECURED LOAN!
BANK!
ACCOUNT!
BANK!
ACCOUNT!
UNSECURED LOAN!
PHONE NUMBER!
CREDIT CARD!
SSN/ ID NUMBER!
PHONE NUMBER!
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
ACCOUNT!
HOLDER!
ADDRESS!
PHONE NUMBER!
$!
APPLICATION!
75. Decisions
$
Better Decisions
Graphs add highly predictive
features to models; adding
accuracy without altering current
workflows
Machine Learning Pipeline
Machine Learning Pipeline
Traditional methods based on ”flat
data” simplify, or leave out entirely,
predictive relationship and network
data