Dr. Jesús Barrasa's slides from his talk at Connected Data London. Jesús, who is a senior field engineer at Neo4j presented how semantic web principles can be used in a graph database.

1. Explicit Semantics in Graph DBs
Driving Digital Transformation With Neo4j
Dr. Jesús Barrasa - Senior Consultant

2. Me: Jesús Barrasa
:Researcher
{
topic: Semantic Tech
PhD Thesis : Mapping
RDBMS-RDF
}
:HOLDS_POSITION
{
from: 2002
to: 2007
}
:HOLDS_POSITION
{
from: 2007
to: 2014
}
:Graph+SemanticsSpecialist
{
sector: Telecoms
useCases: [dependency modelling,
impact analysis, root cause,…]
}
pecialist
cy modelling,
cause,…]
:HOLDS_POSITION
{
from: 2015
to: …
}
:GraphConsultant
{
vendor:Neo4j
useCases: [fraud detection,
recommendation,MDM…]
}
ction,
M…]
:Person
{
name: J. Barrasa
twitter: @BarrasaDV
mail: jesus@neo4j.com
}

3. My talk today
What do they mean when they call it semantics?
Semantics in Graph DBs. Is this a thing?
Explicit semantics in Neo4j: A 5 min experiment
And now what?

4. Quick show of hands: What do you
mean when you say semantics?
Fragment Syntax Semantic?
:JohnSmith :livesIn :London
:London :cityIn :England RDF/Turtle Y/N
{ uri: “JohnSmith”, livesIn: { uri:“London”, cityIn: { uri:”England”}}} JSON Y/N
(j:Resource { uri:’JohnSmith’})-[:livesIn]->(l:Resource { uri:’London’})
(l)-[:cityIn]->(e:Resource { uri:’England’}) Cypher Y/N
<rdf:RDF>
<rdf:Description rdf:about=“Cljkfkojhg”>
<fjoijlgkih rdf:resource=“lksdfjslkjghhhlkjsfd”/>
</rdf:Description>
</rdf:RDF>
RDF/XML Y/N
{"@graph": [{
"@id": "http://places.com/London",
"v:cityIn": { "@id": “http://places.com/England" }
},{
"@id": "v:cityIn",
"rdfs:domain": { "@id": “v:Location"}, "rdfs:range": { "@id": "v:Location"}
}]}
RDFS (RDF/
JSON-LD)
Y/N
| who | livesIn | | what | cityIn |
|JohnSmith | London | | London | England |
CSV Y/N

5. Misconception… that some
cheeky RDF vendors keep alive.
Two facts:
John Smith lives in London and London is a city
in England.
And magic will happen…
Two triples:
:JohnSmith :livesIn :London
:London :cityIn :England
One Question:
Who lives in England?
The Query:
SELECT ?who WHERE { ?who :LivesIn
:England }
huh? what’s going on ?!?

6. A little detail is missing…
If someone lives in a city and that city is in a country, then
we can derive that this someone lives in that country.
But someone has to EXPLICITLY state this for an “intelligent
semantic DB” to apply it to the data
?x :livesIn ?city ^ ?city :cityIn ?ctry => ?x :LivesIn ?ctry
Making the semantics of your data explicit is what operates
the ‘magic’
Now we can try again: Who lives in England?
ANSWER: JohnSmith

7. more “magic”...
Now with a reasoning engine that understands RDFS
semantics we can ask: What locations do we know?
:cityIn rdfs:domain :Location
:cityIn rdfs:range :Location
:cityIn is a relationship stated between two locations
ANSWER: London and England
I’m a purist (and have a PhD in description logics) so I’m not
writing rules, I use a set of primitives with well defined
meaning like InverseFunctional, Domain, Disjoint, Range
and a generic rules engine will apply them for me on my data.

8. back to the origins: The Semantic Web
(1) https://www.w3.org/DesignIssues/RDFnot.html
TBL in 1998: The Semantic Web is not AI
[…] it does not imply some magical artificial
intelligence which allows machines to
comprehend human mumblings. It only indicates
a machine's ability to solve a well-defined
problem by performing well-defined
operations on existing well-defined data.
Instead of asking machines to understand
people's language, it involves asking people to
make the extra effort. (1)

9. :JohnSmith :livesIn :London
:London :cityIn :England
…
:JohnSmith :livesIn :London
:London :cityIn :England
…
An example with (a bit of) code
:JohnSmith :livesIn :London
:London :cityIn :England
…
Implicit Semantics
Application SELECt ?loc WHERE
{ ?loc a :Location}
SELECt ?loc WHERE
{ ?loc a :Location}
Application
Explicit SemanticsSELECt ?loc WHERE
{
}
Application
{ ?loc a :Location }
union
{ [] :livesIn ?loc }
union
{ ?loc :cityIn [] }
union
{ [] :cityIn ?loc }
?x :livesIn ?place
=> ?place a :Location
OWL/RDFS Ontology
:cityIn rdfs:domain :Location
:cityIn rdfs:range :Location
OWL/RDFS Reasoner
?prop refs:domain ?class ^
?res ?prop []
=> ?res a ?class(FC/BC/H) Rules Engine

10. So a semantic DB is…
• A graph database (often based on the RDF model,
but…)
• Some explicit description of the data in the graph
(typically RDFS/OWL or rule based)
• An (often rules based) domain-independent
processor that applies the explicit semantics

11. Two consequences:
No, but don’t worry. It’s a GRAPH!
1. Is my DB still semantic if I just use RDF
but don’t make my semantics explicit?
2. RDF is not the only way to build a
semantic DB?
You got it!

12. After all… RDF Triple store (!)
• Graph based data/knowledge/… exchange model
• Using RDF as an exchange format does not necessarily imply using tiple
storage: Think of Linked Data
• Neo4j for instance can expose graph data as RDF as do other stores or
middleware (DV, D2R).
Indeed, one of the main driving forces for the Semantic
web, has always been the expression, on the Web, of
the vast amount of relational database information in a
way that can be processed by machines (1).
(1) https://www.w3.org/DesignIssues/RDB-RDF.html
RDF is a standard model for data
interchange on the Web.
https://www.w3.org/TR/2004/REC-rdf-primer-20040210/
RDF is a directed, labeled graph data format for
representing information in the Web.
https://www.w3.org/TR/rdf-sparql-query/

13. So…Linked data, semantic data,
graph data…
Graph Data
LPG Data RDF Data
Explicit Semantics
:abc :custId :def({id:’abc’})-[:custId]->(id:’def’)
:custId a owl:InverseFunctionalProperty
:custId rdfs:domain :Customer
:Customer rdfs:subClassOf :Person
({id: ‘custId’})-[:domain]->({id: ‘Customer’})
({id: ‘Customer’})-[:subClassOf]->({id: ‘Person’})
:abc :def
:custId
:abc :def
:custId

14. Build a Semantic Graph DB in 5’
• Learn an Ontology from a data set (3’)
• Formalize the ontology -> make semantics explicit (1’)
• Use these semantics to drive your ‘intelligent’
application (1’)

15. The dataset
• 230K+ article summaries from the Financial Times
Demo...
articleDate articleTitle articleUrl keywords description storySummary
2012-12-17 “Vintage performance” http://… "EU integration, economies,
natural resources, energy policy,
industry, entrepreneurs, investment,
restaurants, filmmaking,
central and eastern Europe”
2013-01-09 "Google and the US
economy”
"On a warm autumn afternoon in Tokaj,
Laszlo Kalocsai waits patiently for his
grapes to turn mouldy. On the edge of the
Carpathian mountains, the best wines are
only possible once the fog-borne fungus
Botrytis cinerea has risen from the
wetlands"
"Thanks to
investment and a
focus on quality
eastern European
wines are now
much in demand”
http://… “Lex” ”A nice easy question: is the US economy
growing or shrinking? The majority view,
to simplify slightly, is that increasing
employment and rising house prices must
amount to an expansion. Dissident
pessimists have a more complicated story.
The current"
"Bullish
investors
should look at
world’s biggest
internet search
group"

16. More on explicit semantics in
Non-RDF graph DBs
https://jesusbarrasa.wordpress.com

17. THANK YOU!