4. 4
Neo4j PS in the real world
Solution Delivery
and Management
• Packaged Services
• Typically 5-25 days
• Neo4j advises
• Customer builds
• 80% of projects
• Custom Scoped
• 50+ man days
• Neo4j delivers
• Customer supports
• 20% of projects
5. PROFESSIONAL
SERVICES
GRAPH ACADEMY
SOLUTIONS
CUSTOMER SUPPORT
● Packaged Services
● Staff Augmentation
● Project/Solution Delivery
● Class room training
● Online/Virtual training
● Certification
● Innovation Labs
● Solution Workshops
● Solutions Development
● 24x7x365 & KB
● Platinum support
● Cloud Managed Services
● DBaaS (NEW)
● Agile Solution Support
Training
Enablement
Solution Delivery
& Management
Organization and offerings
8. • Agility -- constantly changing requirements
• Intuitiveness – so that everybody in your organization can
understand and influence the solution
• High Performance to support connected data scenarios
• Value Connections
• 360 degree views (customer, product, etc.)
• Finding patterns, traversing through the data
• Sysadmin friendliness
• Hardware efficiency8
Technical Requirements
(need of Neo4j Graph Based Solution)
13. Solution (Foundation) Framework
Neo4j Graph Platform
Recom Telco
App
Solution Foundation
Framework
Development &
Administration
Analytics
Tooling
Graph
Analytics
Graph
Transactions
Data Integration
Discovery & VisualizationDrivers
App App
14. Solution (Foundation) Framework
Neo4j Graph Platform
Recom Telco
App
Solution Foundation
Framework
Development &
Administration
Analytics
Tooling
Graph
Analytics
Graph
Transactions
Data Integration
Discovery & VisualizationDrivers
App App
API dev
3rd party
graph viz
Custom dev with
graph viz libraries
3rd party
analytics
Python,
Java ML, ...
Kettle 3rd party DI/EAI
Docker
Kubernetes
Git
Lineage
GRANDstack
26. Where AI and ML fit in
26
Development &
Administration
Analytics
Tooling
Graph
Analytics
Graph
Transactions
Data Integration
Discovery & VisualizationDrivers & APIs
AI
27. Differences between ML and Analytics
28
Machine learning:
• Determine domain parameters
• Historical-based discoveries
• Learn and improve without explicit
programming
28. Graph analytics:
• Uses inherent graph structures
• Uncover real-world networks
through their connections
• Forecast complex network
behavior and identify action
Differences between ML and Analytics
29. (Some) today challenges with Machine Learning:
• Doesn’t take multiple relationship hops into account
• Takes time to iteratively train a model
• Computational inefficiency of connecting data
Machine Learning Pipeline
Benefits of Mixing Graph Analytics with ML
Graphs bring:
• Context to machine learning
• Feature filtration
• Connected feature extraction
30. Neo4j has an ‘out of the box’ Graph Algorithms plugin:
• Pathfinding and Search
• Centrality and Importance
• Community Detection
• Similarity and Machine Learning Workflow
• Link Prediction
Many different ways to work with your ML algorithms in Neo4j:
• Support for many languages (Python, .Net, Java, Go, Ruby, etc.)
• Different data integration options
• Triggers, event-driven architecture, user-defined functions
31
Working with Graph Analytics and ML
31. Pathfinding
& Search
Centrality /
Importance
Community
Detection
Similarity
GraphConnect 2017 GraphConnect 2018
• Minimum Weight Spanning Tree
• Shortest Path
• Single-Source Shortest Path
• All Pairs Shortest Path
• PageRank
• Article Rank
• Betweenness Centrality
• Closeness Centrality
• Louvain
• Label Propagation
• Connected Components
• Harmonic Centrality
• Eigenvector Centrality
• Degree Centrality
• A* Shortest Path
• Yen’s K Shortest Path
• Random Walk
• Jaccard Similarity
• Cosine Similarity
• Pearson Similarity
• Strongly Connected
Components
• Triangle Count /
• Clustering Coefficient
• Balanced Triads
Machine Learning and Graph Algorithms in
Neo4j
• Euclidean Distance
• Overlap Similarity
Link
Prediction
• Adamic Adar
• Common Neighbors
• Preferential Attachment
• Resource Allocation
• Same Community
• Total Neighbors
32
neo4j.com/
graph-algorithms-
book/
32. IoT/Connected Home:
• Master Data Management
• Entity resolution using
community detection and similarity
Customer Experience Management:
• Customer journey path analysis
(path finding)
33
Graph Analytics and Algorithm Examples
33
33. Knowledge graph example:
• Using topic finding ML processes
e.g. Latent Dirichlet Allocation (LDA)
• Feeding the output into a graph database
• Search for topics, find related concepts, etc.
34
Graph and Machine Learning Examples
Recommendation engine example:
• Use ML processes such as collaborative filtering
• Enrich graph with the output
• Use graph as feedback for future iterations
36. Customer Use Case:
• Leading online platform to showcase and discover creative work
• More than 10 million members
• Allows creatives to share their work with millions of daily visitors
• Highlights Adobe software used in the creation process
• Drives people to the Adobe Creative Cloud
• Social platform for discovery, learning, and more
38
Adobe – Project Behance
Activity feed:
• Mongo (2011) - 125 nodes, dataset size of about 20tb
(terabytes)
• Cassandra (2015) - 48 nodes, dataset size of about 50tb
(terabytes)
• Neo4j (2018) - 3 nodes, dataset size of 40gb (gigabytes)
5 day Bootcamp
37. 39
Large Commercial Bank
Customer Journey
Innovation
Lab
Staff Augmentation
Campaign Management
Innovation
Lab
Fraud Project
80 person days
TBD
Another
Innovation Lab
38. 40
Conclusion
• Neo4j Professional Services makes customer projects successful
through:
• Enablement
• Project / solution delivery
• Graph Based Solutions as accelerators
• Neo4j is the foundation for AI and ML
• Customers are using Neo4j to drive success and deliver value
40. Our Neo4j activity implementation has led to a great decrease in complexity, storage, and
infrastructure costs. Our full dataset size is now around 40 GB, down from 50 TB of data
that we had stored in Cassandra. We’re able to power our entire activity feed infrastructure
using a cluster of 3 Neo4j instances, down from 48 Cassandra instances of pretty much
equal specs. That has also led to reduced infrastructure costs. Most importantly, it’s been
a breeze for our operations staff to manage since the architecture is simple and lean.”
David Fox, Adobe, Oct 2018
42
Customer Quote
How can Neo4j Services help you to get there?