Optimizing Your Supply Chain with Neo4j
Dr. Michael Moore, Senior Director, Strategy and Innovation, Neo4j
With the world’s supply chain system in crisis, it’s clear that better solutions are needed. Digital twins built on knowledge graph technology allow you to achieve an end-to-end view of the process, supporting real-time monitoring of critical assets.
2. Michael Moore, Ph.D.
Senior Director, Strategy & Innovation
michael.moore@neo4j.com
Optimizing Your Supply Chain
with Neo4j
Neo4j Inc. All rights reserved 2023
2
3. Consider: What Drives Your Processes?
It’s not the numbers, it’s the relationships behind them
Plants
Warehouses
Suppliers
Distributors
Competitors
Partners
Regulations
Employees
Citizens
Customers
Products
Parts
Services
Regions
3 Neo4j Inc. All rights reserved 2023
3
4. This is a Graph
Neo4j Inc. All rights reserved 2023
4
5. “By 2025, graph technologies will be
used in 80% of data and analytics
innovations...”
Top 10 Trends in Data and Analytics, 11 May 2020, Rita Sallam et al.
5 Neo4j Inc. All rights reserved 2023
5
6. :Supplier :Part
:Product
Nodes
represent entities
(typically nouns)
Relationships are directional
(but can be queried in either direction)
Relationships connect nodes
(typically verbs)
Relationships can have any
number of properties
(key/value pairs)
Nodes can have any
number of properties
(key/value pairs)
name: Amy Peters
date_of_birth: 1984-03-01
customer_id: 1
:PRODUCED
mfr_dt: 2018-01-20
:HAS_PART
Nodes can have
zero or more labels
(role/set)
:
P
U
R
C
H
A
S
E
D
p
o
_
d
a
t
e
:
2
0
1
8
-
0
1
-
2
6
The Property Graph Data Model
6
:Buyer
:HAS_SUB_PART
Relationships can
be recursive
(unbounded
hierarchies/paths)
Neo4j Inc. All rights reserved 2023
6
7. Depict the business
as a graph
Squash the graph
into tables
Jam in foreign keys to
relate the records,
populate global index
Cheap Memory Makes Graphs Compelling
SQL RDBMS workarounds to conserve memory
1979: Oracle v2.0 Released (yes, 43 years ago!)
= hidden technical debt
per MB
per MB
7 Neo4j Inc. All rights reserved 2023
7
https://jcmit.net/memoryprice.htm
8. Graphs Have Low Complexity and High Fidelity
SQL RDBMS ER Diagram Graph (“Whiteboard”)
8 Neo4j Inc. All rights reserved 2023
8
9. Graph Query Performance vs Data Size
Connectedness and Size of Data Set
Response
Time
Relational and
other NoSQL
databases
Native Graph Database
1000x Advantage
Minutes to milliseconds
5+ hops
3+ degrees
Thousands of connections
0 to 2 hops
0 to 3 degrees
Few connections
9
A native graph system with index-free adjacency does not have to move
through any other type of data structures to find links between the nodes.
Neo4j Inc. All rights reserved 2023
9
10. Graph Transactions,
Storage & Querying
Graph Analytics, ML,
& Data Science
Intelligent Operational Systems Better Predictions for Analytics
Use of Graph Database
as Data Warehouse
Internal use for
Identity &
Management
Gen AI/LLM +
Knowledge
Graph
Neo4j Database Built for Operational and Analytical Workloads
10 Neo4j Inc. All rights reserved 2023
10
11. Enterprise
Trust &
Security
Runs on Cloud
(Azure, Google and AWS)
and on-premises
Scale: Autonomous Clustering
& Composite Databases (Fabric)
Hybrid
Workloads
Native Graph
Architecture
Powers
Graph Data
Science
Comprehensive
Toolset &
Ecosystem
connectivity
Large
Community
growing 80%
yoy
Supports all
data shapes &
relationships
Neo4j Database Built for Operational and Analytical Workloads
11 Neo4j Inc. All rights reserved 2023
11
12. Graphs Naturally Handle Complex Data
A B C D E
A B C D E
One-to-Many
Relationships
Across Many
Entities
Wide Data Complex Data Hierarchical & Recursive Data
Many-to-Many
Relationships
Nested Tree
Structures
Recursion
(Self-Joins)
Deep Paths
Link Prediction
(If C relates to A and A relates to E,
then C must relate to E)
Vector Similarity
Hidden Data
Legacy Data Frozen Data
Legacy SQL Systems Data Lake Fact Tables Graph Data Science - Machine Reasoning
A
C
E
Neo4j Inc. All rights reserved 2023
12
13. Demand Shaping Inventory Positioning Perfect Fulfillment Sustainability
Consider a omni-channel,
multi-echelon supply chain and
demand & supply risk to
determine optimized inventory
levels
Combine distributed order
management with warehouse
and transportation optimization
and enable automation of
processes
Reduce energy consumption
across operations, measure
risks, measure gas emissions,
adopt a circular economy and
measure the social impact of
your supply chain
Understand customers and
market trends to dynamically
adjust segmentation, promotions
and generate highly accurate
forecasts
Demand Shaping Inventory Positioning Perfect Fulfillment Sustainability
Supply Chain Complexity Issues
End-to-End Supply Chain Visibility
Neo4j Inc. All rights reserved 2023
13
14. Scope 3 Requires Upstream and Downstream Reporting
https://www.epa.gov/climateleadership/scope-3-inventory-guidance
Neo4j Inc. All rights reserved 2023
14
15. GHG Reporting Requirements
• March 21, 2022:
SEC released new proposals
for climate-related risk
disclosures.
• February 2024:
First disclosures on Scope 1
and 2 for large organizations.
• February 2025: Disclosures on
Scope 3 emissions and
emissions intensity required
for large organizations.
Organizations' supply chains often account for more than
90% of their greenhouse gas (GHG) emissions, when taking
into account their overall climate impacts.
Neo4j Inc. All rights reserved 2023
15
16. Formidable Data Collection Requirements
Upstream Value Chain Data
x Emission Factors
+
Downstream Value Chain Data
x Emission Factors
+
Existing Scope 1 and Scope
Estimates
= Total Carbon Estimate
Neo4j Inc. All rights reserved 2023
16
18. Supply Chains Are Graphs
• Global Supply Chain Visibility
• Routing, Logistics, Distribution
• BOM Management
• Supply Chain Resiliency
• Scope 3 Carbon Reporting
Supply chains have complex linkages
and deep hierarchies and are most
naturally modeled as a graph.
Neo4j Inc. All rights reserved 2023
18
19. OrbitMI
Maritime Routing
• Digital twin PLM system with full BoM
for all Army equipment, including costs,
armaments, force posture and readiness.
• Complex analysis is 7.5 X faster
• Rapid “What-If” analysis enables more
agile response to global scenarios
U.S. Army
Force Readiness
• Knowledge graph of 27 Million warranty
& service documents
• Graph AI learns failure mode “prime
examples” to anticipate maintenance
• Improves equipment lifespan and
customer satisfaction
Caterpillar
AI for Maintenance
Customer Examples of Supply Chain Twins
• Digital twin of global maritime routes
• Subsecond route planning
• Global carbon emissions reduced by
60,000 tons annually
• $12-16M ROI for OrbitMI customers
Neo4j Inc. All rights reserved 2023
19
20. Supply Chain Optimization using Neo4j
Neo4j Graph Data
Science Library
Neo4j
Database
Neo4j
Bloom
Inference & Predictions Graph Digital Twin Visualization & Investigation
Neo4j Inc. All rights reserved 2023
20
21. Supply Chain Digital Twins
Digital Supply Chain Twins—Conceptual Clarification, Use Cases and Benefits
Benno Gerlach, Simon Zarnitz, Benjamin Nitsche and Frank Straube
Logistics 2021, 5, 86. https://doi.org/10.3390/logistics5040086
Network Level
Site Level
Neo4j Inc. All rights reserved 2023
21
22. Constructing the Knowledge Graph
• Ontologies
• Taxonomies
• Friendly Naming
• Schema/Structure
• Master Data
• Slowly Changing Dims
• Hierarchies
• Mappings
• Business Processes
• Signal Events
• Granular Detail
• Real Time Context
• Communities
• Dependencies
• Isomorphic Subgraphs
• ML Predictions
A knowledge graph combines consistent business semantics, entities extracted
and unified from source data, detailed transactional flows, and in-graph
analytics/inference for decision support.
Semantic
Conventions
Resolved
Entities
Operational
Transactions
Graph
Inference
Neo4j Inc. All rights reserved 2023
22
23. Neo4j Inc. All rights reserved 2023
23
Supply Chain Graph Algorithms
Pathfinding
Shortest, least
expensive routes,
alternate routes.
Centrality
Quickly locate critical
bottlenecks or facilities with high
activity.
Node Similarity
Identify replacement
parts/facilities to handle
inventory shortfalls and
disruptions
Embeddings
Enable semantic search
and Generative AI
experiences
24. North American Rail Network (NARN) Digital Twin
North American Rail Network
Total 527K Miles of Track
USDOT NARN GEOJSON datasets - nodes and lines
253K nodes, 715K relationships, 1GB
https://github.com/graphadvantage/neo4j-na-rail-network
Norfolk Southern rail network graph plotted using NeoMap
Neo4j Inc. All rights reserved 2023
24
25. Neo4j Graph Database
Yards connected by track type
Graph Data Science-Projecting Network Views
main-lines-network
Has low bridges & tunnels
double-stack-network
No low bridges & tunnels
Neo4j Graph Data Science
Projected virtual graph views
standard rail car
double-stack
intermodal rail car
Neo4j Inc. All rights reserved 2023
25
26. Neo4j Pathfinding
//Project main lines
CALL gds.graph.project(
'main-lines-network', 'Node', {relType: {type:
'CONNECTS_MIO', orientation: 'UNDIRECTED',
properties: {miles: {property: 'miles',
defaultValue: 1}}}}, {})
//Project double stack lines
CALL gds.graph.project(
'double-stack-network', 'Node', {relType: {type:
'CONNECTS_DS', orientation: 'UNDIRECTED',
properties: {miles: {property: 'miles',
defaultValue: 1}}}}, {})
//Find a route
CALL gds.shortestPath.dijkstra.stream(
'main-lines-network',
{relationshipWeightProperty: 'MILES',
sourceNode: id(start), targetNode: id(end)})
What is the lowest cost path between two points?
“Cost” could be physical distance, fuel consumption, fees, carbon emissions, etc
Dijkstra shortest path solution visualized in NeoDash
Minimize
this cost
Neo4j Inc. All rights reserved 2023
26
27. Google Supply Chain Twin L2 Graph Data Model
Neo4j Inc. All rights reserved 2023
27
28. Neo4j Inc. All rights reserved 2023
The Modern Supply Chain is a Knowledge Graph
28 https://www2.deloitte.com/content/dam/insights/us/articles/3465_Digital-supply-network/DUP_Digital-supply-network.pdf