A comparison of two graph data models - RDF and Property Graphs.

1. Graph Data
Linked Data and
Property Graphs

2. Contents
● Example
● LD/RDF
● PG
● SPARQL
● Gremlin/Cypher
● BIG data
– Apache Giraph / Bulk Synchronous Processing
● Ideally?
– +arrays+URIs+attributes

3. Graphs
● A set of vertices (nodes) and edges (arcs)
● Except the useful kind have labels on edges
● … and the nodes are just dots.

4. Graphs
G = ( V , E )
V – Vertexes (Nodes)
E – Edges (Arcs, Links)

5. Graphs

6. Graphs For Information
Alice
Bob
Eve
listensTo
knows

7. Graphs For Information
Alice
... has a name “Alice Hacker”
... has an employee number

8. Linked Data / RDF
● Standards
– What it means
– Syntaxes for exchanging data
– Query language
● URI name things globally
● Uniform representation
– Link to another thing is same as link to a value
● Complex structures encoded in the basic mechanism
● “Schemaless” data integration

9. Linked Data
★ make your stuff available on the Web (whatever
format) under an open license
★★ make it available as structured data (e.g., Excel
instead of image scan of a table)
★★★ use non-proprietary formats (e.g., CSV instead
of Excel)
★★★★ use URIs to denote things, so that people
can point at your stuff
★★★★★ link your data to other data to provide context
http://5stardata.info/

10. Linked Data / RDF
http://example/alice
"Alice Hacker"
foaf:name
http://example/bob
foaf:knows
prefix person: <http://example/person/>
prefix foaf: <http://xmlns.com/foaf/0.1/>
<http://example/alice>
foaf:name "Alice Hacker" ;
foaf:knows <http://example/bob> .
<http://example/bob>
foaf:name "Bob Tester" ;
foaf:knows <http://example/alice> .
foaf:name
Bob Tester
foaf:knows

11. JSON-LD
● Links and semantics for the JSON ecosystem
{
"@context" : "http://example/person.jsonld",
"@graph" : [ {
"@id" : "http://example/alice",
"knows" : "http://example/bob",
"name" : "Alice Hacker"
}, {
"@id" : "http://example/bob",
"knows" : "http://example/alice",
"name" : "Bob Tester"
} ]
}

12. SPARQL Query
prefix person: <http://example/person/>
prefix foaf: <http://xmlns.com/foaf/0.1/>
<http://example/alice> foaf:name "Alice Hacker" ;
foaf:knows <http://example/bob> .
<http://example/bob> foaf:name "Bob Tester" ;
foaf:knows <http://example/alice> .
PREFIX foaf: <http://xmlns.com/foaf/0.1/>
SELECT ?name
WHERE
{ ?person foaf:name "Alice Hacker" ;
?person foaf:knows ?name . }
----------------
| name |
================
| "Bob Tester" |
----------------

13. Property Graphs
● Separates Links and Attributes
● Nodes have attributes
– … and so do edges
● Different definitions
– https://github.com/tinkerpop/ is the de facto standard
– Not universal
● Data exchange (web publishing) is not an objective
● Analysis and schema-less data applications

14. Build
Graph graph = new TinkerGraph();
Vertex a = graph.addVertex("alice");
Vertex b = graph.addVertex("bob");
a.setProperty("name","Alice Hacker");
b.setProperty("name","Bob Coder");
Edge e1 = graph.addEdge("k1", a, b, "knows");
Edge e2 = graph.addEdge("k2", b, a, "knows") ;

15. GSON
{
"edges" : [
{
"_id" : "k1" ,
"_inV" : "bob" ,
"_label" : "knows" ,
"_outV" : "alice" ,
"_type" : "edge"
} ,
{
"_id" : "k2" ,
"_inV" : "alice" ,
"_label" : "knows" ,
"_outV" : "bob" ,
"_type" : "edge"
}
] ,
"mode" : "NORMAL" ,
"vertices" : [
{
"_id" : "bob" ,
"_type" : "vertex" ,
"name" : "Bob Coder"
} ,
{
"_id" : "alice" ,
"_type" : "vertex" ,
"name" : "Alice Hacker"
}

16. Gremlin
// Groovy to Java
@SuppressWarnings("unchecked")
Pipe<Vertex,Vertex> pipe = Gremlin.compile("g.v('alice').out('knows').name");
for(Object name : pipe) {
System.out.println((String) name);
}
g.v('alice').out('knows').name

17. Cypher Query
● Neo4J specific
● Property Graph + “labels” (= types) – node names
CREATE (alice { name: 'Alice Hacker'} ) ,
(bob { name: 'Bob Tester'} ) ,
(alice) -[:knows]-> (bob) ,
(bob) -[:knows]-> (alice)
MATCH (a)-[:knows]->x
WHERE a.name = 'Alice Hacker'
RETURN x.name