�ݺ�ߣ

Neo4j
Graph Database
for Ruby and Rails
Pablo Delgado
Madrid, Conferencia Rails 2010
Wednesday, November 10, 2010

Presentation Titel | Author | City, 11/05/2010 2
GRAPHS
Extracted from
“The Graph Traversal Programming Pattern”
Rodriguez, Marko A.
Graph Systems Architect at AT&T Interactive

Undirected Graph
• Vertices
All Vertices denote the
same type of node
• Edges
- All edges denote the
same type of relationship
- All edges denote
symmetric relationship

Directed Graph
• Vertices
All Vertices denote the
same type of node
• Edges
- All edges denote the
same type of relationship
- All edges denote
asymmetric relationship

Single-Relational Graph
Definition
• Without a way to differentiate types of edges, all
edges have the same meaning, hence single-
relational graph
Limitations
• Are only able to represent a single kind of vertex
(user, city, etc)
• Are only able to represent a single kind of edge
(follows, citation, friends)

Multi-Relational Graph
Definition
• With the labeling of the edges we obtain the
possibility to express different relationships (follows,
likes, include,etc)
Gain
• Edges can now have different meanings and vertex
can have different types.
• follows: user -> user
• likes: user -> webpage
• cites: article -> paper
likes PageUser

Increased expressivity
with multi-relational
graphs
Multi-Relational Graph
likes
likes likes
likes
likes
likes
follows
follows
follows

Property Graph
Definition
• A property graph extends a multi-relational graph
allowing both, vertices and edges to maintain a
key/value property map.
• Properties are useful to express non-relational data
• This allows adding more meaning to an edge
“John Smith liked http://twitter.com on 2010/10/01”
likes PageUser
name = John Smith url = http://twiiter.com
date = 2010/10/01

Increased meaning on
edges and nodes in a
property graph
John liked http://twittpic.com on 2010/06/06
Xurde liked http://yumit.com on 2010/03/03
Ivan liked http://uwi.sh on 2010/07/07
Property Graph
http://twittpic.com
John
Paul
2010/06/06
http://facebook.com
http://google.com
http://yumit.com
Xurde
http://uwi.sh/
Ivan
http://panoramio.com
2010/03/03
2007/03/01
2010/7/07
2005/04/05
2005/02/05

GRAPH DATABASE

Graph with Relational Database
B
A
D
C
outV | inV
-----+-----
A | B
A | C
C | D
D | A

Graph with JSON Database
B
A
D
C
{
A: {
out : [B,C], in : [D]
},
B: {
in : [A]
},
C: {
out : [D], in : [A]
},
D: {
out : [A], in : [C]
}
}

Graph with XML Database
B
A
D
C
<graphml>
<graph>
<node id=A />
<node id=B />
<node id=C />
<node id=C />
<edge source=A target=B />
<edge source=A target=C />
<edge source=C target=D />
<edge source=D target=A />
</graph>
</graphml>

Graph Database
Definition
• Every element (i.e. vertex or edge) has a direct
pointer to its adjacent element.
• No O(log2(n)) index lookup required to determine
which vertex is adjacent to which other vertex.
• If the graph is connected, the graph as a whole is a
single atomic data structure.

Graph Databases and
Index-free Adjacency
B
A
D
C
• In a graph database, vertex A has direct references to its adjacent
vertices.
• Constant time cost to move from A to B and C . It is only dependent
upon the number of edges from vertex A (local lookup).
The Graph (explicit)

Non-Graph Databases and
Index-Based Adjacency
B
A
D
C
• In a non-graph database, an index lookup is needed to determine
what is adjacent to A.
• log2(n) time cost to move to B and C. It is dependent upon the total
number of vertices and edges in the database (global lookup).
A B C D
[B,C] [D] [A,B][ ]
The Graph (implicit)The Index (explicit)

Neo4j

Neo4j is a Graph Database
• Non-relational, transactional (ACID), embedded
• Data is stored as a Graph / Network
‣Nodes and relationships with properties
‣“Property Graph” or “edge-labeled multi-digraph”
• Schema free, bottom-up data model design
Neo4j is Open Source / Free Software
• AGPLv3
• Commercial (“dual license”) license available
What is Neo4j?

• Traversal APIs
Neo4j core traversers
Blueprint Pipes (TinkerPop Productions)
• SPARQL - “SQL for linked data” - query by graph pattern
matching
SELECT ?person WHERE {
?person neo4j:KNOWS ?friend .
?friend neo4j:KNOWS ?foe .
?foe neo4j:name "Larry Ellison"
}
Find all persons that KNOWS a friend that KNOWS someone named “Larry Ellison”.
• Gremlin - “perl for graphs” - query by traversal
./outE[@label='KNOWS']/inV[@age > 30]/@name
Names of all People I know with age > 30.
Neo4j Query Languages

Data Manipulation API
GraphDatabaseService graphDb = getGraphDbInstanceSomehow();
// Create Extractor Cobb
Node cobb = graphDb.createNode();
cobb.setProperty( "name", "Cobb" );
cobb.setProperty( "role", "Extractor" );
// Create Ariadne Architect
Node ariadne = graphDb.createNode();
ariadne.setProperty( "name", "Ariadne" );
ariadne.setProperty( "role", "Architect" );
// Create relationship representing they know each other
cobb.createRelationshipTo( ariadne, RelTypes.KNOWS );
// ... similarly add more relations here

Data Manipulation API
GraphDatabaseService graphDb = getGraphDbInstanceSomehow();
Transaction tx = graphDb.beginTx();
try {
// Create Extractor Cobb
Node cobb = graphDb.createNode();
cobb.setProperty( "name", "Cobb" );
cobb.setProperty( "role", "Extractor" );
// Create Ariadne Architect
Node ariadne = graphDb.createNode();
ariadne.setProperty( "name", "Ariadne" );
ariadne.setProperty( "role", "Architect" );
// Create relationship representing they know each other
cobb.createRelationshipTo( ariadne, RelTypes.KNOWS );
// ... similarly add more relations here
tx.success();
} finally {
tx.finsh();
}

Neo4j
with
Ruby and Rails

• neo4j.rb - by Andreas Ronge
https://github.com/andreasronge/neo4j
A Graph Database for JRuby. It uses the java lib neo4j as
storage and Lucene for quering/indexing
• neo4jr-simple by Matthew Deiters
http://github.com/mdeiters/neo4jr-simple
Its a JRuby wrapper for Neo4j
Neo4j for Ruby and Rails

Inception
since=
6years
activity = undisclosed
Knows relationship

Inception
Hire relationship

Inception
In Love relationship

Inception
since=
6years
activity = undisclosed
Knows relationship
Hire relationship
In Love relationship

Inception with Neo4j
Neo4j::Transaction.run do
cobb = Node.new :name => "Cobb", :role => "Extractor"
ariadne = Node.new :name => "Ariadne", :role => "Architect", location => "Paris"
saito = Node.new :name => "Saito", :role => "Employer", :rich => true
mal = Node.new :name => "Mal", :role => "Wife",
arthur = Node.new :name => "Arthur", :role => "Partner"
eames = Node.new :name => "Eames", :role => "Forger", :location => "Mombasa"
yusuf = Node.new :name => "Yusuf", :role => "Chemist", :phone => "+254 41 787878"
fischer = Node.new :name => "Fisher Jr.", :role => "Target", :age => 29
Relationship.new(:root, Neo4j.ref_node, cobb)
Relationship.new(:knows, cobb, arthur)[:since] => "6 years"
Relationship.new(:knows, cobb, eames)
Relationship.new(:knows, arthur, eames)
Relationship.new(:knows, cobb, mal)
Relationship.new(:in_love, cobb, mal)
Relationship.new(:knows, arthur, mal)
Relationship.new(:knows, eames, yusuf)[:activity] => "undisclosed"
Relationship.new(:knows, saito, fischer)
Relationship.new(:hire, cobb, ariadne)
Relationship.new(:hire, cobb, yusuf)
Relationship.new(:hire, cobb, eames)
Relationship.new(:hire, saito, cobb)
Relationship.new(:hire, saito, arthur)
end
cobb.outgoing(:knows).depth(:all).filter { self[:name] == "Yusuf" }.first

• ACID Atomicity, Consistency, Isolation, Durability
Only write locks, no read locks
Transactions
# DO STUFF
end

Object Oriented Mapping
class Person
include Neo4j::NodeMixin
property :name
end
person = Person.new
person.name = "pablo"
end

Lucene in Neo4j.rb
class Person
property :name
index :name
end
person = Person.new
person.name = "pablo"
end
Person.find("name: pablo")

Traversals in Neo4j.rb
cobb.outgoing(:knows).depth(:all).filter { self[:name] == "Yusuf" }
pablo.outgoing(:follows).depth(2).filter { self[:age] == 30 }
From all people who cobb knows, and they know is there anyone namedYusuf?
From all people who pablo follows, which ones are 30 years old?

Relations
pablete = Node.new(:login => "pablete")
mauro = Node.new(:login => "malditogeek")
tony = Node.new(:login => "antoniogarrote")
pablete.outgoing(:friends) << mauro << tony
pablete.outgoing(:friends).each {|friend| puts friend[:login]}
end

Relations with NodeMixin
class Person
property :name
has_n :friends
end
pablete = Person.new(:login => "pablete")
mauro = Person.new(:login => "malditogeek")
tony = Person.new(:login => "antoniogarrote")
pablete.friends << mauro << tony
pablete.friends.each {|friend| puts friend[:login]}
end

Rails Drop-in Replacement
class User < ActiveRecord::Base
attr_accessor :password
attr_accessible :name, :email, :password,:password_confirmation
after_save :encrypt_password
validates :name, :presence => true,
:length => { :maximum => 50}
validates :password, :presence => true,
:confirmation => true,
:length => {:withing => 6..40}
has_one :profile
private
def encrypt_password
self.salt = make_salt if new_record?
self.encrypted_password = encrypt(password)
end
end

Rails Drop-in Replacement
class User < Neo4j::Model
attr_accessor :password
attr_accessible :name, :email, :password,:password_confirmation
after_save :encrypt_password
validates :name, :presence => true,
:length => { :maximum => 50}
validates :password, :presence => true,
:confirmation => true,
:length => {:withing => 6..40}
property :name, :email, :salt, :encrypted_password
index :email
has_one :profile
private
def encrypt_password
self.salt = make_salt if new_record?
self.encrypted_password = encrypt(password)
end
end

DEMO TIME

Gephi - Makes Graph Handy
Github Example:
5 Users:
-pablete
-martincik
-ppeszko
-antoniogarrote
-malditogeek
Users in RED
Repositories in BLUE
http:://github.com/pablete/conferenciarails2010

Special thanks to:
Andreas Ronge
http://twitter.com/ronge
Peter Neubauer
http://twitter.com/peterneubauer
Marko A. Rodriguez
http://twitter.com/twarko

Thanks!
Pablo Delgado
pablete@gmail.com
@pablete

�ݺ�ߣ

Neo4j for Ruby and Rails

More Related Content

Neo4j for Ruby and Rails