Presentation for the WebTuesday http://webtuesday.ch/meetings/20121211/ Dec 11th, 2012.

1. Semantic Web
Dr. Sebastian Ryszard Kruk
Dec 11th, 2012; Web Tuesday, Zurich

2. Outline
✤ Quick walk through the history of the Web
✤ The core principles of the Semantic Web
✤ Some vocabularies, schemas and ontologies you should know
✤ Publishing the Web 3.0 way
✤ Consuming semantics
✤ Semantic Web gotchas
✤ Summary: how Semantic Web technologies can help

3. The history of the Web

4. Let’s go back in time
✤ 1937 - H.G. Wells: “World Brain”
✤ 1945 - Vannevar Bush: Memex
✤ 1969 - Doug Engelbart: ARPANET, online collaboration, mouse
✤ 1989 - sir Tim Berners-Lee: WWW
✤ 1995 - Dublin Core: semantics from librarian world
✤ 1999 - prof. Stefan Decker: OntoBroker
✤ 2000 - DARPA publish DAML specification
✤ 2001 - TBL, Handler i Lassila: “The Semantic Web” (SciAm)
✤ 2009 - TBL - 20th anniversary: “we’ve got all the tools"
✤ 2009 - Google introduces Rich Snippets/RDFa; BestBuy sales
increase
✤ 2010 - Facebook starts Open Graph
✤ 2012 - Google announces Knowledge Graph

5. The blueprints of the Web

6. The core principles

7. The “roadmap” cake

8. RDF is not a format!

9. RDF is not a format!
Leather made by
Wood Chair
is a made
made of is a
Carpenter
Man
sits on
is a
has
Back
Leg
count used for
4 Sitting

10. RDF is not a format!
Leather made by
Wood Chair
is a made
predicate
subject
made of object is a
Carpenter
a statement Man
sits on
is a
has
Back
Leg
count used for
4 Sitting

11. Resource, Literals and BNodes
✤ Resource - everything identified by URI/IRI
✤ Literal - a value (“four”)
✤ with an optional type (4^^xsd:int)
✤ or optional lang tag (“four”@en)
✤ BNode - point in the graph with only local _:name
✤ Statement - a triple (subject, predicate, object)
✤ Reification - a statement about the statement

12. Resource, Literals and BNodes
✤ Resource - everything identified by URI/IRI
✤ Literalns: a value (“four”)
- <http://test.test/ns> .
@prefix
@prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> .
@prefix xsd: <http://www.w3.org/2001/XMLSchema> .
✤ with an optional type (4^^xsd:int)
ns:Chair ns:has ns:Leg.
_:sentence a rdf:Statement; tag (“four”@en)
or optional lang
✤
rdf:subject ns:Chair;
rdf:predicate ns:has;
✤ BNoderdf:object ns:Leg;the graph with only
- point in local _:name
ns:count “4”^^xsd:int.
✤ Statement - a triple (subject, predicate, object)
✤ Reification - a statement about the statement

13. Vocabularies
✤ Schemes:
✤ Taxonomy: narrower/broader
✤ Thesaurus: related
✤ Folksonomy: “machines are (using) us” Peter Mika [Yahoo]
✤ Ontology: “An ontology is a specification of a conceptualization” Tom
Gruber, 1993 [*Siri]

14. Ontology languages
✤ RDF Schema: classes, properties, domains and ranges
✤ OWL (Web Ontology Language) - Lite, DL and Full, and OWL 2: (among
others) inverse functional property, symmetry, transitiveness and
disjointness
✤ But also: UML, ERD, ...

15. Vocabularies and ontologies

16. Dublin Core
✤ One of the oldest, originates from digital libraries
✤ Goals:
✤ semantic and syntactic interoperability
✤ shareable descriptions, vocabularies and constrains
✤ Defines:
✤ 15 basic concepts, such as dc:title, dc:creator
✤ Additional classes and concepts in DC Qualifiers

17. FOAF+SIOC+SKOS
http://sioc-project.org

18. Good Relations
✤ De-facto standard for semantic e-commerce
✤ Supported by Google via Rich Snippets
✤ BestBuy showed how to increase revenue with it
✤ Part of the schema.org specification
✤ Example information:
✤ the geo-location of the store
✤ the countries the store ships to
✤ the payment options the ship accepts
✤ Distribute, open “Amazon”

19. schema.org
✤ Vocabulary for Rich Snippets and Knowledge Graph
✤ Yahoo+Bing+Google
✤ Microdata vs RDFa (Lite)

20. Publishing semantics

21. Linked (Open) Data
http://linkeddata.org

22. 5-Star Data Rules
✤ URIs = names for things
✤ HTTP URIs = enable look up
✤ Provide useful information
upon lookup, using the
standards (RDF*, SPARQL)
✤ Link to other URIs = enable
discovery of more things.
http://www.w3.org/DesignIssues/LinkedData.html

23. 5-Star Data Rules
✤ URIs = names for things
✤ HTTP URIs = enable look up
✤ Provide useful information
upon lookup, using the
standards (RDF*, SPARQL)
✤ Link to other URIs = enable
discovery of more things.
http://www.w3.org/DesignIssues/LinkedData.html

24. Content negotiation

25. RDFa and RDFa Lite
✤ Extension to XHTML/HTML5 for embeding RDF graph semantics in
HTML syntax:
✤ property vs rel
✤ resource vs about
✤ typeof
✤ vocab, prefix

26. RDFa and RDFa Lite
✤ Extension to XHTML/HTML5 for embeding RDF graph semantics in
HTML syntax:
<p vocab="http://schema.org/"
property vs rel
✤ prefix="ov: http://open.vocab.org/terms/"
resource="#skruk"
typeof="Person">
My name is <span property="name">Sebastian Kruk</span>
✤ resource vs about
and you can give me a ring via
<span property="telephone">+41-78-87-961-89</span>.
<img property="image" src="http://bit.ly/QAYF10" />
My favorite animal is the <span property="ov:preferredAnimal">Hamster</span>.
✤
</p>typeof
✤ vocab, prefix

27. Rich Snippets (microdata)
<div class="col-right small" itemid="http://example/" itemtype="http://schema.org/
FurnitureStore">
<ul class="list"><li><div itemprop="address" itemscope="" itemtype="http://schema.org/
PostalAddress">
<p>
<span itemprop="postalCode">5210</span> <span itemprop="addressLocality">Windisch</
span><br>
<span itemprop="addressCountry">Switzerland</span><br>
</p>
<p>
<a class="email" href="mailto:windisch@kruk.me" itemprop="email">windisch@kruk.me</a>
<span itemprop="telephone">+41788796189</span><br>
</p>
</div></li></ul>

28. Rich Snippets (RDFa Lite)
<div class="col-right small" resource="http://example.com/" typeof="FurnitureStore"
vocab="http://schema.org/">
<ul class="list"><li><div property="address" typeof="PostalAddress">
<p>
<span property="postalCode">5210</span> <span property="addressLocality">Windisch</
span><br>
<span property="addressCountry">Switzerland</span><br>
</p>
<p>
<a class="email" href="mailto:windisch@kruk.me" property="email">windisch@kruk.me</a>
<span property="telephone">+41788796189</span><br>
</p>
</div></li></ul>
</div>

29. Consuming semantics

30. SPARQL
✤ W3C standard for querying RDF graphs
✤ Purposely made similar to SQL (to the extent possible)
✤ SPARQL endpoint: a standard way to expose (open) data published
by the service

31. Path queries
✤ ?a :d “D”.
✤ ?a :a ?b. ?b :e “E”; :f “F”.
✤ ?a :b [ :c “C” ].

32. Path queries
:c
C
:b
✤ ?a :d “D”.
:a :e
?a ?b E
✤ ?a :a ?b. ?b :e “E”; :f “F”.
:d :f
✤ ?a :b [ :c “C” ].
D F

33. SELECT and more
✤ PREFIX dc: <http://purl.org/dc/elements/1.1/>
✤ SELECT ?title
WHERE { <http://mybook.com> dc:title ?title.}
✤ CONSTRUCT { ?b ns:isKnownBy ?a }
WHERE { ?a foaf:knows ?b }
✤ ASK { <http://kruk.me/> foaf:knows ?c }
✤ DESCRIBE ?a
WHERE { ?a foaf:knows <http://kruk.me> }

34. SPARQL Filtering
✤ OPTIONAL { ?x ns:y ?z }
✤ FILTER ( xsd:dateTime(?date) >
xsd:dateTime("2010-03-01T00:00:00Z") )
✤ FILTER regex( ?value, "This is a S+", "i" )
✤ isURI/isIRI, isBlank, isLiteral, sameTerm, langMatches
✤ str, lang, datatype

35. Python RDFLib
from rdflib.graph import Graph, URIRef
from rdflib.namespace import Namespace
from rdflib import plugin
plugin.register('sparql', rdflib.query.Processor,
'rdfextras.sparql.processor', 'Processor')
plugin.register('sparql', rdflib.query.Result,
'rdfextras.sparql.query', 'SPARQLQueryResult')
plugin.register('turtle', rdflib.serializer.Serializer,
'rdflib.plugins.serializers.turtle', 'TurtleSerializer')
namespaces = { 'skos' : Namespace('http://www.w3.org/2004/02/skos/core#'), }
def list_friends(f_name, uri):
g = Graph()
g.parse(f_name, format="n3")
query = "SELECT ?fn WHERE { ?f foaf:knows <%s>; foaf:name ?fn. }" % uri
return [ f for f in g.query(query, initNs=namespaces) ]

36. Read RDFa into RDF
from pyRdfa import pyRdfa, Options
def rdf2turtle(url):
options = Options(output_processor_graph=True)
turtle = pyRdfa(options=options).rdf_from_source(url, outputFormat='turtle')
return turtle

37. Gotchas

38. Don’t overcomplicate
✤ Gödel’s Incompleteness Theorems
✤ First-Order Logic vs OWL-Full / CYC
✤ Jim Handler: “Little semantics go a long way”
✤ Folksonomies

39. OOP vs Description Logic
✤ B rdfs:subClassOf A
C rdfs:domain A
D rdfs:domain B
✤ Infers:
✤ C rdfs:domain B [WRONG!]
✤ D rdfs:domain A [OK]

40. OOP vs Description Logic
✤ B rdfs:subClassOf A
C rdfs:domain A
D rdfs:domain B domain
A C
✤ Infers: subClassOf
domain
B D
✤ C rdfs:domain B [WRONG!]
✤ D rdfs:domain A [OK]

41. Open World Assumptions
✤ RDF is by it’s virtue a distributed being
✤ You might only have a part of it
✤ The rest is on Linked Data
✤ Unlike in the close world assumption reasoning, if there is no
statement “Sebastian isFrom Poland” it does not mean, I am not.

42. RDBMS and NoSQL vs RDF
✤ We can store RDF in RDBMS; even easier in NoSQL
✤ Certain SPARQL path queries can be translated to SQL
✤ Usually no joint queries NoSQL ➡ DIY path querying

43. How can web semantics help
✤ interoperability
✤ information retrieval
✤ recommendations and matchmaking
✤ SEO

44. Linked Data is King
Dr. Sebastian Ryszard Kruk
CTO/CRO Knowledge Hives sp. z o.o., sebastian@kruk.me