Mark Maslyn presented an introduction to Neo4j, an open-source graph database, covering its applications, advantages, and Java API syntax. The presentation emphasized the utility of graph databases in various fields like social media and fraud detection, demonstrating their natural representation of data and faster query responses. A live demo of an animal guessing game using Neo4j showcased its decision tree capabilities.

1. NEO4J OPEN SOURCE GRAPH
DATABASE
Presented by Mark Maslyn – mmaslyn@msn.com to
Denver Open Source User Group (4/7/15) and
Graph Nerds of Boulder (5/14/15)

2. THREE PARTS TO THIS NEO4J PRESENTATION
• What Are Graph Databases and Why Are They
Useful
• Neo4J Application Modes and Java API Syntax
• Demo Animal Guessing Game using Neo4J
Decision Tree

3. WHAT ARE GRAPH DATABASES ?
• Graph Databases are a type of NoSQL ( non-
relational) database
• In a Graph Database entities (nouns) are
represented as nodes
• Relationships are represented by edges
connecting nodes
• Both nodes and relationships can have properties

4. FRIEND-OF EXAMPLE
FRIEND-OFFRIEND-OF
Properties:
Name: Mark M
Language: Java
Properties:
Name: Tom M
Language: Java
Properties:
Name: Tom F
Language: Scala

5. SOME COMPANIES AND INDUSTRIES WITH GRAPH
DATABASE APPLICATIONS
• Social Media (Facebook, LinkedIn, Twitter)
• Configuration Management (Assimilation
Systems)
• Retail Recommendation (Walmart)
• Resource Authorization (Telenor)
• Fraud Detection ( Banks and Credit Card
Companies)
• Online Education (Pearson)
• Bioinformatics (Bio4j)

6. WHERE GRAPH DATABASES ARE
ADVANTAGEOUS
• Can provide a more natural representation
for the data
• Can give a faster query response
• Recommendation engines

7. EXAMPLES WHERE A GRAPH REPRESENTATION
IS A NATURAL MODEL

8. FLIGHT ROUTE MAP – SF TO NY
3
4
1
2
1.5
2
3
1
1
3
1.5
FLYS-TO
FLYS-TO FLYS-TO
FLYS-TO
FLYS-TO
FLYS-TOFLYS-TO
FLYS-TO
FLYS-TO
FLYS-TO

9. TREE OF LIFE
http://oceanexplorer.noaa.gov/okeanos after Woese (1990)

10. TWO LEVEL PROTEIN NETWORK DIAGRAM
From CYTOSCAPE.ORG
Two Level Tree
p53

11. EXAMPLE OF QUERY SPEED-UP

12. FRIEND-OF-A-FRIEND-OF-A-FRIEND …
From Vukotic, et al (2014)

13. EXECUTION TIME RDBMS VS GRAPH DATABASE FOR
DIFFERENT DEPTHS OF FRIEND SEARCH (1000 FRIENDS)
Depth Execution Time (sec) Count Result
2 0.028 ~900
3 0.213 ~999
4 10.273 ~999
5 92.613 ~999
Depth Execution Time (sec) Count Result
2 0.04 ~900
3 0.06 ~999
4 0.07 ~999
5 0.07 ~999
MYSQL RDMS
NEO4J
From Vukotic, et al (2014)

14. EXAMPLES OF RECOMMENDATIONS BASED
ON GRAPH CONNECTIONS AND CONTENT

15. TYPICAL APPROACHES TO GENERATING
RECOMMENDATIONS (AND PREDICTIONS)
• COLLABORATIVE FILTERING – Recommendations based on
common attributes between you and / or your connections –
Online dating model
• CONTENT BASED FILTERING - Deriving a second level of
information from the data and use that to derive
recommendations – Content classification model
• GRAPH TOPOLOGY – Infer links based on graph topology.

16. COLLABORATIVE FILTERING RECOMMENDATION BASED ON
FRIENDS CHOICES – I WILL LIKE WHAT MY FRIENDS LIKE
FRIEND-OFFRIEND-OF
Properties:
Name: Mark M
Language: Java
Properties:
Name: Tom M
Language: Java, Scala
Movie: Empire Strikes Back
Properties:
Name: Tom F
Language: Scala, Java
Movie: Raiders of the Lost Ark

17. CONTENT BASED RECOMMENDATIONS BASED ON
CLASSIFICATION OF FRIENDS MOVIE SELECTION
FRIEND-OFFRIEND-OF
Properties:
Name: Mark M
Language: Java
Properties:
Name: Tom M
Language: Java, Scala
Movie: Empire Strikes Back
Properties:
Name: Tom F
Language: Scala, Java
Movie: Raiders of the Lost Ark

18. CONTENT CLASSIFICATION BASED ON TERMS
SPORTS CATEGORY
FOOTBALL BASEBALL
Broncos
Seahawks
Ronnie HillmanPeyton Manning
Rockies
1 .. N SPORTS
Russell Wilson
CATEGORY
SPORT
TEAM
PLAYER
1 .. N TEAMS
1 .. N PLAYERS

19. FINDING RELATIONSHIPS BASED ON GRAPH TOPOLOGY
FRIEND-OFFRIEND-OF
Transitive Relationship
A FRIEND OF B, B FRIEND OF C, Therefore High
Probability A FRIEND OF C

20. CONCLUSION: A TRIADIC CLOSURE OR MY FRIENDS ARE
LIKELY TO BE FRIENDS WITH EACH OTHER
FRIEND-OFFRIEND-OF
Properties:
Name: Mark M
Language: Java
Properties:
Name: Tom M
Language: Java
Properties:
Name: Tom F
Language: Scala
TRIADIC CLOSURE

21. FRIEND RELATIONSHIPS BETWEEN PARTNERS IN A
KARATE CLUB – BEFORE SPLIT INTO TWO CLUBS
From Zachary (1977)

22. KARATE CLUBS MEMBERSHIP AFTER SPLIT
From Zachary (1977)

23. PART II – NEO4J GRAPH DATABASE MODES
AND JAVA API SYNTAX

24. NEO4J GRAPH DATABASE MODES
• Embedded Mode Uses Local NEO4J Jar Files
• Server Mode Uses RESTFul API’s
• Browser Client Mode Using CYPHER Query
Language

25. EMBEDDED MODE WITH JAVA API – OPENING
A GRAPH DATABASE
private static GraphDatabaseService graphDb;
private static String ANIMAL_DATABASE_LOC =
"/home/ubuntu/animal_game/animal.db";
public AnimalGame() {
// open the graph database
graphDb = new GraphDatabaseFactory().newEmbeddedDatabase(
ANIMAL_DATABASE_LOC);
// enable clean db shutdown on Ctrl-C
registerShutdownHook( graphDb );
}

26. JAVA API – CREATING AN INDEX
// create index using a combination of the node label and
the name property
try (Transaction tx = graphDb.beginTx()) {
graphDb.schema().indexFor(NodeLabels.PERSON).on(
"name").create();
tx.success();
}

27. JAVA API – CREATING A DATABASE NODE AND
SETTING PROPERTY VALUES
// create node
Node node = graphDb.createNode();
// add label to identify what group the node belongs to
node.addLabel(nodeLabel);
// set property values
node.setProperty(AnimalConstants.NAME, name);
node.setProperty(AnimalConstants.TEXT, text);
node.setProperty(AnimalConstants.TYPE, nodeType);
// make a list of possible answers (answerList)
// these match with relationships
node.setProperty(AnimalConstants.ANSWERS, answerList);

28. JAVA API – CREATING RELATIONSHIPS
BETWEEN NODES
public enum RelTypes implements RelationshipType {
YES,
NO
}
public void createRelationship(Node node1,
Node node2,
RelTypes rt) {
// relationship between node1 and node2
// direction is from node1 to node2
node1.createRelationshipTo((Node) node2, rt);
}

29. JAVA API - QUERYING THE DATABASE FOR A
NODE BY LABEL AND NAME
public Node getNodeByLabelAndName(String nodeName) {
Node node = null;
ResourceIterable<Node>nodeList =
graphDb.findNodesByLabelAndProperty(
NodeLabels.PERSON, "name", nodeName);
If (nodeList != null) {
try (Transaction tx = graphDb.beginTx()) {
for ( Node nodeL : nodeList )
node = nodeL;
tx.success();
}
}
return node;
}

30. NEO4J SERVER RESTFUL API
• Neo4J Server Provides Fine-Grained REST Calls
• Returns JSON Content Type
• Higher Level Libraries Available for Java, .NET, Python,
etc. to Wrap the Lower Level Calls

31. NEO4J SERVER RESTFUL API REQUEST /
RESPONSE WITH DETAILED AVAILABLE URL’S
RESPONSE
{
"extensions": {},
"node": "http://localhost:7474/db/data/node",
"node_index": "http://localhost:7474/db/data/index/node",
"relationship_index": "http://localhost:7474/db/data/index/relationship",
"extensions_info": "http://localhost:7474/db/data/ext",
"relationship_types": "http://localhost:7474/db/data/relationship/types",
"batch": "http://localhost:7474/db/data/batch",
"cypher": "http://localhost:7474/db/data/cypher",
"indexes": "http://localhost:7474/db/data/schema/index",
"constraints": "http://localhost:7474/db/data/schema/constraint",
"transaction": "http://localhost:7474/db/data/transaction",
"node_labels": "http://localhost:7474/db/data/labels",
"neo4j_version": "2.1.7"
}
REQUEST (GET): http://localhost:7474/db/data

32. NEO4J SERVER RESTFUL API – DRILLING DOWN FOR
META-DATA FROM ANIMAL DATABASE
REQUEST: http://localhost:7474/db/data/relationship/types
JSON RESPONSE:
[
"YES",
"NO"
]

33. RETRIEVING A SINGLE NODE AND ITS PROPERTIES
REQUEST: http://localhost:7474/db/data/node/0
JSON RESPONSE:
{
…..
"metadata": {
"id": 0,
"labels": [
“ANIMALS"
]
},
"data": {
"text": "Does the animal live on land ?",
"name": "start",
"answers": [
"Y/YES",
"N/NO",
"Q/QUIT"
],
"type": "question"
}
}

34. RETRIEVING A NODE USING THE RESTFUL LIBRARY
FROM JAVA CODE
private static GraphDatabaseService graphDb;
private static String ANIMAL_DATABASE_URL =
“http://localhost:7474/db/data”;
// public constructor
public AnimalGame() {
// connect to the graph database server
graphDb = new RestGraphDatabase(
ANIMAL_DATABASE_URL);
// wrap the request and retrieve the node
Node startNode =
graphDb.findNodesByLabelAndProperty(
NODE_LABEL,
“name”,
“start”);

35. NEO4J BROWSER BASED WEB ADMIN CLIENT
• Browser Connects to Neo4J Server.
• Graphically Displays Database Nodes and Relationships in a
Dashboard. Can Drill Down on Each Node.
• Allows User to Execute Cypher Queries on Database From
the Browser.

36. NEO4J BROWSER WEB ADMIN DASHBOARD

37. CLICK ON A NODE FOR MORE INFO

38. CYPHER QUERY LANGUAGE
• Cypher is Neo4J’s version of SQL
• Queries can be entered programmatically in Java
or through the browser interface
• Queries can retrieve nodes, relationships,
property values or call functions such as return
“shortest path traversal”
• Declarative syntax using “Ascii Art”

39. NODES - CYPHER SYNTAX
Cypher uses “Ascii Art” syntax that reflects graph elements. Nodes
are represented by parentheses “( )”, relationships by arrows “-->”.
The arrow head indicates the direction of the relationship
Adding constraints “find two Person group nodes a and b that are
connected by a relationship”.
Match (a) –-> (b)
Return a.name, b.name
Match (a:Person) --> (b:Person)
Return a.name, b.name

40. RELATIONSHIPS - CYPHER SYNTAX
Anonymous relationships match all relationship and are indicated
by the arrow alone
This query syntax is used to find two nodes a and b that are
connected by the specific “ACTED_IN” relationship to match
actors with movies
Match (a) --> (b)
Match (a) –[:ACTED_IN]-> (b)

41. PROPERTY GRAPH EXAMPLE FOR A MOVIE
DATABASE

42. SIX DEGREES OF KEVIN BACON GAME – HOW
MANY LINKS ARE REQUIRED TO CONNECT AN
ACTOR TO KEVIN BACON ?
Kevin Bacon (left) and Tom Hanks in Apollo 13 (1995)

43. CYPHER QUERY FOR SIX DEGREES OF KEVIN BACON
FOR MEG RYAN USING NEO4J SHORTEST PATH
FUNCTION
MATCH p=shortestPath(
(b:Person {name:"Kevin Bacon"})-[*]-(m:Person {name:"Meg
Ryan"})
)
RETURN p

44. AND THE ANSWER IS …
Meg Ryan’s “Bacon” Number is 2

45. NEO4J DECISION TREE – ANIMAL GUESSING GAME
Mammal ?
Has Stripes ? Slithers ?
Does it Growl ?
Has Trunk ?
Zebra ?Tiger ?
Elephant ?
Snake?
YES
YES
YES
YES
YES
NO
NO
NO

46. PART III
DEMO ANIMAL GUESSING GAME – AUDIENCE
VOLUNTEER REQUESTED

47. A Final Comment on Connected Graphs
(Paraphrasing) Imagination is the relationship
connecting two nodes…

48. TWO BOOKS I RECOMMEND FOR MORE
INFORMATION ON GRAPH THEORY AND NEO4J
Graph Theory Neo4J
By Jennifer Golbeck, 2013 By Aleksa Vukotic, et al 2014