An introduction to Semantic Web and Linked Data

Semantic Web and Linked Data or how to link data and schemas on the web a W3C tutorial by Fabien Gandon, http://fabien.info, @fabien_gandon WWW 2014

semantic web mentioned by Tim BL in 1994 at WWW [Tim Berners-Lee 1994, http://www.w3.org/Talks/WWW94Tim/]

machines don’t. we identify and interpret information,

IRI HTML HTTP identification address communication WEB

W3C®SEMANTIC WEB STANDARD STACK

RDFstands for Resource: pages, dogs, ideas... everything that can have a URI Description: attributes, features, and relations of the resources Framework: model, languages and syntaxes for these descriptions

RDFis a triple model i.e. every piece of knowledge is broken down into ( subject , predicate , object )

doc.html has for author Fabien and has for theme Music

doc.html has for author Fabien doc.html has for theme Music

( doc.html , author , Fabien ) ( doc.html , theme , Music ) ( subject , predicate , object )

Predicate Subject Object a triplethe RDF atom

RDFis also a graph model to link the descriptions of resources

RDFtriples can be seen as arcs of a graph (vertex,edge,vertex)

( doc.html , author , Fabien ) ( doc.html , theme , Music )

Fabien author doc.html theme Music

identify what exists on the web http://my-site.fr identify, on the web, what exists http://animals.org/this-zebra

http://ns.inria.fr/fabien.gandon#me http://inria.fr/schema#author http://inria.fr/rr/doc.html http://inria.fr/schema#theme Music

open and link data in a global giant graph

RDFin values of properties can also be literals i.e. strings of characters

( doc.html , author , Fabien ) ( doc.html , theme , "Music" )

http://ns.inria.fr/fabien.gandon#me http://inria.fr/schema#author http://inria.fr/rr/doc.html http://inria.fr/schema#theme "Music"

http://ns.inria.fr/fabien.gandon#me http://inria.fr/schema#author Music http://inria.fr/rr/doc.html http://inria.fr/rr/doc.html http://inria.fr/schema#theme

<rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22- rdf-syntax-ns#" xmlns:inria="http://inria.fr/schema#" > <rdf:Description rdf:about="http://inria.fr/rr/doc.html"> <inria:author rdf:resource= "http://ns.inria.fr/fabien.gandon#me"/> <inria:theme>Music</inria:theme> </rdf:Description> </rdf:RDF>

RDFhas other syntaxes (Turtle, TriG, N-Triples, N-Quads, JSON, RDFa)

Turtle @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> . @prefix inria: <http://inria.fr/schema#> . <http://inria.fr/rr/doc.html> inria:author <http://ns.inria.fr/fabien.gandon#me> ; inria:theme "Music" .

N-Triples <http://inria.fr/rr/doc.html> <http://inria.fr/schema#author> <http://ns.inria.fr/fabien.gandon#me> . <http://inria.fr/rr/doc.html> <http://inria.fr/schema#theme> "Music" .

writing rules for RDF triples • the subject is always a resource (never a literal) • properties are binary relations and their types are identified by IRIs • the value is a resource or a literal

blank nodes (bnodes) http://bu.ch/l23.html author "My Life" title "John" surname "Doe" firstname handy anonymous nodes (existential quantification) there exist a resource such that… {  r ; …} <rdf:Description rdf:about="http://bu.ch/123.html "> <author> <rdf:Description> <surname>Doe</surname> <firstname>John</firstname> </rdf:Description> </author> <title>My Life</title> </rdf:Description> <http://bu.ch/123.html> author [surname "Doe" ; firstname "John" . ] ; title "My Life" .

XML schema datatypes & literals standard literals are xsd:string type literals with datatypes from XML Schema <rdf:Description rdf:about="#Fabien"> <teaching rdf:datatype="http://www.w3.org/2001/XMLSchema#boolean"> true</teaching> <birth rdf:datatype="http://www.w3.org/2001/XMLSchema#date"> 1975-07-31</birth> </rdf:Description/> #Fabien teaching "true"^^xsd:boolean ; birth "1975-07-31"^^xsd:date . #Fabien "true"^^xsd:boolean "1975-07-31"^^xsd:date teaching birth

XML Schema datatypes W3C-http://www.w3.org/TR/xmlschema-2/

langue <Book> <title xml:lang=‘fr’>Seigneur des anneaux</title> <title xml:lang=‘en’>Lord of the rings</title> </Book> <Book> title "Seigneur des anneaux"@fr ; title "Lord of the rings"@en . literals with languages and without are disjoint “Fabien”  “Fabien”@en  “Fabien”@fr

typing resources using URIs to identify the types <urn://~fgandon> rdf:type <http://www.inria.fr/schema#Person> a resource can have several types <urn://~fgandon> rdf:type <http://www.inria.fr/schema#Person> <urn://~fgandon> rdf:type <http://www.inria.fr/schema#Researcher> <urn://~fgandon> rdf:type <http://www.mit.edu/schema#Lecturer> <rdf:Description rdf:about="urn://~fgandon"> <rdf:type rdf:resource="http://www.inria.fr/schema#Person" /> <name>Fabien</name> </rdf:Description> <in:Person rdf:about="urn://~fgandon"> <name>Fabien</name> </in:Person> <urn://~fgandon> a in:Person ; name "Fabien" .

question: <?xml version="1.0"?> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:exs="http://example.org/schema#"> <rdf:Description rdf:about="http://example.org/doc.html"> <rdf:type rdf:resource="http://example.org/schema#Report"/> <exs:theme rdf:resource="http://example.org#Music"/> <exs:theme rdf:resource="http://example.org#History"/> <exs:nbPages rdf:datatype="http://www.w3.org/2001/XMLSchema#int">23</exs:nbPages> </rdf:Description> </rdf:RDF> meaning ?

question: <?xml version="1.0"?> <rdf:RDF xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:exs="http://example.org/schema#"> <rdf:Description rdf:about="http://example.org/doc.html"> <rdf:type rdf:resource="http://example.org/schema#Report"/> <exs:theme rdf:resource="http://example.org#Music"/> <exs:theme rdf:resource="http://example.org#History"/> <exs:nbPages rdf:datatype="http://www.w3.org/2001/XMLSchema#int">23</exs:nbPages> </rdf:Description> </rdf:RDF> exs:Report rdf:type exs:nbPages “23”^^xsd:int exs:theme http://example.org/doc.html http://example.org#Music http://example.org#History exs:theme

bags = unordered groups <rdf:Description rdf:about="#"> <author> <rdf:Bag> <rdf:li>Ivan Herman</rdf:li> <rdf:li>Fabien Gandon</rdf:li> </rdf:Bag> </author> </rdf:Description> <#> author _:a _:a rdf:_1 “Ivan Herman” _:a rdf:_2 “Fabien Gandon” <#> author [ a rdf:Bag ; rdf:li "Ivan Herman" ; rdf:li "Fabien Gandon" . ] .

sequence ordered group of resources or literals <rdf:Description rdf:about="#partition"> <contains> <rdf:Seq> <rdf:li rdf:about="#C"/> <rdf:li rdf:about="#C"/> <rdf:li rdf:about="#C"/> <rdf:li rdf:about="#D"/> <rdf:li rdf:about="#E"/> </rdf:Seq> </contains> </rdf:Description> <partition> contains [ a rdf:Seq ; rdf:li "C" ; rdf:li "C" ; rdf:li "C" ; rdf:li "D" ; rdf:li "E" . ] .

alternativese.g. title of a book in different languages <rdf:Description rdf:about="#book"> <title> <rdf:Alt> <rdf:li xml:lang="fr">l’homme qui prenait sa femme pour un chapeau</rdf:li> <rdf:li xml:lang="en">the man who mistook his wife for a hat</rdf:li> </rdf:Alt> </title> </rdf:Description> <#book> title [ a rdf:Alt ; rdf:li "l’homme…"@fr ; rdf:li "the man…"@en . ] .

collectionexhaustive and ordered list <rdf:Description rdf:about="#week"> <dividedIn rdf:parseType="Collection"> <rdf:Description rdf:about="#monday"/> <rdf:Description rdf:about="#tuesday"/> <rdf:Description rdf:about="#wednesday"/> <rdf:Description rdf:about="#thursday"/> <rdf:Description rdf:about="#friday"/> <rdf:Description rdf:about="#saturday"/> <rdf:Description rdf:about="#sunday"/> </devidedIn> </rdf:Description> wednesday friday sunday nil monday tuesday thursday saturday firstrest List _:a _:b _:c _:d _:e _:f _:g <#week> dividedIn ( <#monday> <#tuesday> <#wednesday> <#thursday> <#friday> <#saturday> <#sunday> ) .

RDF(named) graphs group triples in graphs named by IRIs

http://ns.inria.fr/fabien.gandon#me http://inria.fr/schema#author Music http://inria.fr/rr/doc.html http://inria.fr/rr/doc.html http://inria.fr/schema#theme http://inria.fr/people http://inria.fr/topics

TriG @prefix rdf: <http://www.w3.org/1999/02/22-rdf-syntax-ns#> . @prefix inria: <http://inria.fr/schema#> . GRAPH <http://inria.fr/people> { <http://inria.fr/rr/doc.html> inria:author <http://ns.inria.fr/fabien.gandon#me> . } GRAPH <http://inria.fr/topics> { <http://inria.fr/rr/doc.html> inria:theme "Music" . }

N-Quads <http://inria.fr/rr/doc.html> <http://inria.fr/schema#author> <http://ns.inria.fr/fabien.gandon#me> <http://inria.fr/people> . <http://inria.fr/rr/doc.html> <http://inria.fr/schema#theme> "Music" <http://inria.fr/topics> .

rdf:about rdf:type ex:ingredients rdf:label dc:creator ex:weight

openmodel • extensible vocabulary based on URIs • anyone can say anything about anything http://my_domain.org/my_path/my_type

May 2007 April 2008 September 2008 March 2009 September 2010 Linking Open Data Linking Open Data cloud diagram, by Richard Cyganiak and Anja Jentzsch. http://lod-cloud.net/ September 2011 0 100 200 300 400 10/10/2006 28/04/2007 14/11/2007 01/06/2008 18/12/2008 06/07/2009 22/01/2010 10/08/2010 26/02/2011 14/09/2011 01/04/2012

thematic content Domains Number of datasets Number of Triples % Out links % Media 25 1 841 852 061 5,82 % 50 440 705 10,01 % Geography 31 6145 532 484 19,43 % 35 812 328 7,11 % Government 49 13 315 009 400 42,09 % 19 343 519 3,84 % Publications 87 2 950 720 693 9,33 % 139 925 218 27,76 % Inter-domain 41 4 184 635 715 13,23 % 63 183 065 12,54 % Life Sciences 41 3 036 336 004 9,60 % 191 844 090 38,06 % Users’ content 20 134 127 413 0,42 % 3 449 143 0,68 % 295 31 634 213 770 503 998 829 42% 20% 13% 10% 9% 6% 0% Government Geography Inter-domain Life Sciences Publications Media Users' content

linked data principles Use RDF as data format  Use HTTP URIs as names for things so that people can look up those names  When someone looks up a URI, provide useful information (RDF, HTML, etc.) using content negotiation  Include links to other URIs so that related things can be discovered HTTP URI GET HTML,RDF,… GET 303

DNShe who controls the name controls the access ex. bit.ly & Libya .fr * .inria isicil

query with SPARQL SPARQL Protocol and RDF Query Language

SPARQL in 3 parts part 1: query language part 2: result format part 3: access protocol

SPARQL query SELECT ... FROM ... WHERE { ... }

examplepersons at least 18-year old PREFIX ex: <http://inria.fr/schema#> SELECT ?person ?name WHERE { ?person rdf:type ex:Person . ?person ex:name ?name . ?person ex:age ?age . FILTER (?age > 17) }

left left x * z left(x,y) left(y,z) right(z,v) right(z,u) right(u,v) left(x,?p) left(?p,z)  right x y z u v right left left

graph mapping / projection classical three clauses: – Select: clause to select the values to be returned – Where: triple/graph pattern to match – Filter: constraints expressed using test functions (XPath 2.0 or external)

SPARQL triples • triples and question marks for variables: ?x rdf:type ex:Person • graph patterns to match: SELECT ?subject ?proprerty ?value WHERE {?subject ?proprerty ?value} • a pattern is, by default, a conjunction of triples SELECT ?x WHERE { ?x rdf:type ex:Person . ?x ex:name ?name . }

question: • Query: SELECT ?name WHERE { ?x name ?name . ?x email ?email . } • Base: _:a name "Fabien" _:b name "Thomas" _:c name "Lincoln" _:d name "Aline" _:b email <mailto:thom@chaka.sn> _:a email <mailto:Fabien.Gandon@inria.fr> _:d email <mailto:avalandre@pachinko.jp> _:a email <mailto:bafien@fabien.info> • Results ? x2

prefixes to use namespaces: PREFIX mit: <http://www.mit.edu#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> SELECT ?student WHERE { ?student mit:registeredAt ?x . ?x foaf:homepage <http://www.mit.edu> . } Base namespace : BASE <…>

SPARQL result failure/ success values found

result formats • a binding i.e. list of all the selected values (SELECT) for each answer found; (stable XML format ; e.g. for XSLT transformations) • RDF sub-graphs for each answer found (RDF/XML format ; e.g. for application integration) • JSON (eg. ajax web applications) • CSV/TSV (eg. export)

example of binding results for previous query in XML <?xml version="1.0"?> <sparql xmlns="http://www.w3.org/2005/sparql-results#"> <head> <variable name="student"/> </head> <results ordered="false" distinct="false"> <result> <binding name="student"> <uri>http//www.mit.edu/data.rdf#ndieng</uri></binding> </result> <result> <binding name="student"> <uri>http//www.mit.edu/data.rdf#jdoe</uri></binding> </result> </sparql>

simplified syntax triples with a common subject: SELECT ?name ?fname WHERE { ?x a Person; name ?name ; firstname ?fname ; author ?y . } list of values ?x firstname "Fabien", "Lucien" . blank node [firstname "Fabien"] or [] firstname "Fabien" SELECT ?name ?fname WHERE { ?x rdf:type Person . ?x name ?name . ?x firstname ?fname . ?x author ?y . }

source PREFIX mit: <http://www.mit.edu#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> FROM http//www.mit.edu/data.rdf SELECT ?student WHERE { ?student mit:registeredAt ?x . ?x foaf:homepage <http://www.mit.edu> . }

optional part PREFIX mit: <http://www.mit.edu#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> SELECT ?student ?name WHERE { ?student mit:registeredAt ?x . ?x foaf:homepage <http://www.mit.edu> . OPTIONAL {? student foaf:name ?name . } } possibly unbound

union alternative graph patterns PREFIX mit: <http://www.mit.edu#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> SELECT ?student ?name WHERE { ?student mit:registeredAt ?x . { { ?x foaf:homepage <http://www.mit.edu> . } UNION { ?x foaf:homepage <www.stanford.edu/> . } } }

sort, filter and limit answers PREFIX mit: <http://www.mit.edu#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> SELECT ?student ?name WHERE { ?student mit:registeredAt ?x . ?x foaf:homepage <http://www.mit.edu> . ?student foaf:name ?name . ? student foaf:age ?age . FILTER (?age > 22) } ORDER BY ?name LIMIT 20 OFFSET 20 students older than 22 years sorted by name results from number #21 to #40

operators • Inside the FILTER: – Comparators: <, >, =, <=, >=, != – Tests on variables : isURI(?x), isBlank(?x), isLiteral(?x), bound(?x) – Regular expression regex(?x, "A.*") – Attributes and values: lang(), datatype(), str() – Casting: xsd:integer(?x) – External functions and extensions – Boolean combinations: &&, || • In the where WHERE: @fr , ^^xsd:integer • In the SELECT: distinct

other functions (v 1.1) isNumeric(Val) test it is a numeric value coalesce(val,…, val) first valid value IRI(Str)/URI(Str) to build an iri/uri from a string BNODE(ID) to build a blank node RAND() random value between 0 and 1 ABS(Val) absolute value CEIL(Val), FLOOR(Val), ROUND(Val) NOW() today’s date DAY(Date), HOURS(Date), MINUTES(Date), MONTH(Date), SECONDS(Date), TIMEZONE(Date), TZ(Date), YEAR(Date) to access different parts of a date MD5(Val), SHA1(Val), SHA256(Val), SHA384(Val), SHA512(Val) hash functions

string / literal functions (v1.1) STRDT(value, type) build a typed literal STRLANG(value, lang) build a literal with a language CONCAT(lit1,…,litn) concatenate a list of literal CONTAINS(lit1,lit2), STRSTARTS(lit1,lit2), STRENDS(lit1,lit2) to test string inclusion SUBSTR(lit, start [,length]) extract a sub string ENCODE_FOR_URI (Str) encodes a string as URI UCASE (Str), LCASE (Str) uppercase and lowercase STRLEN (Str) length of the string

aggregates group by + count, sum, min, max, avg, group_concat, or sample ex. average scores, grouped by the subject, but only where the mean is greater than 10 SELECT (AVG(?score) AS ?average) WHERE { ?student score ?score . } GROUP BY ?student HAVING(AVG(?score) > 10)

question: PREFIX ex: <http://www.exemple.abc#> SELECT ?person WHERE { ?person rdf:type ?type . FILTER(! ( ?type = ex:Man )) }

minussubstract a pattern PREFIX ex: <http://www.exemple.abc#> SELECT ?person WHERE { { ?x rdf:type ex:Person } minus {?x rdf:type ex:Man} }

not existcheck the absence of a pattern PREFIX ex: <http://www.exemple.abc#> SELECT ?person WHERE { ?x ex:memberOf ?org . filter (not exists {?y ex:memberOf <Hell>}) }

if… then… else prefix foaf: <http://xmlns.com/foaf/0.1/> select * where { ?x foaf:name ?name ; foaf:age ?age . filter ( if (langMatches( lang(?name), "FR"), ?age>=18, ?age>=21) ) }

test a value is in / not in a list prefix foaf: <http://xmlns.com/foaf/0.1/> select * where { ?x foaf:name ?n . filter (?n in ("fabien", "olivier", "catherine") ) }

valuespre-defined bindings select ?person where { ?person name ?name . VALUES (?name) { "Peter" "Pedro" "Pierre" } }

paths prefix foaf: <http://xmlns.com/foaf/0.1/> select ?friends_fab where { ?x foaf:name "Fabien Gandon" ; foaf:knows+ ?friends_fab ; } / : sequence | : alternative + : one or several * : zero or several ? : optional ^ : reverse ! : negation {min,max} : length

select expression select ?x (year(?date) as ?year) where { ?x birthdate ?date . }

subquery / nested query select ?name where { {select (max(?age) as ?max) where { ?person age ?age } } ?senior age ?max ?senior name ?name }

construct RDF as result PREFIX mit: <http://www.mit.edu#> PREFIX corp: <http://mycorp.com/schema#> CONSTRUCT { ?student rdf:type corp:FuturExecutive . } WHERE { ?student rdf:type mit:Student . }

free description PREFIX mit: <http://www.mit.edu#> DESCRIBE ?student { ?student rdf:type mit:Student . } or DESCRIBE <…URI…>

SPARQL protocol exchange queries and their results through the web

publication process demo • one-click setup • import raw data • transform to RDF • publish on the web • query online

Test on DBpedia • Connect to: http://dbpedia.org/snorql/ or http://fr.dbpedia.org/sparql or … http://wiki.dbpedia.org/Internationalization/Chapters • Query: SELECT * WHERE { ?x rdfs:label "Paris"@fr . ?x ?p ?v . } LIMIT 10

Linked Data Platform HTTP access to LD resources & containers get, post, put, delete resources from LD servers. GET /people/fab HTTP/1.1 Host: data.inria.fr PUT http://data.inria.fr/people/fab HTTP/1.1 Host: data.inria.fr Content-Type: text/turtle <fab> a foaf:Person ; rdfs:label "Fabien" ; foaf:mbox <fabien.gandon@inria.fr> . ? !

semantic web: linked data and semantics of schemas a little semantics in a world of links

had typed links… the original web

publish the data schemas 180°C+ = ?  + = 

what is the last document you read?

your answer relies on a shared ontology we infer from it we all understood

Document Book Novel Short Story sub type

sub type #12 #21 #47 #48 "document" "book" "livre" "novel" "roman" "short story" "nouvelle" #21  #12 #48  #21#47  #21

#21  #12 #48  #21#47  #21 ontological knowledge formalized #12 #21 #47 #48

languages to formalize ontologies

W3C® PUBLISH SEMANTICS OF SCHEMAS

RDFS provides primitives to Write lightweight ontologies

RDFS to define classes of resources and organize their hierarchy Document Report

RDFS to define relations between resources, their signature and organize their hierarchy creator author Document Person

FO  R  GF  GRmapping modulo an ontology car vehicle car(x)vehicle(x) GF GRvehicle car O

an old schema of RDFS W3C http://www.w3.org/TR/2000/CR-rdf-schema-20000327/

example of RDFS schema <rdf:RDF xml:base ="http://inria.fr/2005/humans.rdfs" xmlns:rdf ="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns ="http://www.w3.org/2000/01/rdf-schema#> <Class rdf:ID="Man"> <subClassOf rdf:resource="#Person"/> <subClassOf rdf:resource="#Male"/> <label xml:lang="en">man</label> <comment xml:lang="en">an adult male person</comment> </Class> <Man> a Class ; subClassOf <Person>, <Male> .

example of RDFS properties <rdf:RDF xml:base ="http://inria.fr/2005/humans.rdfs" xmlns:rdf ="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns ="http://www.w3.org/2000/01/rdf-schema#> <rdf:Property rdf:ID="hasMother"> <subPropertyOf rdf:resource="#hasParent"/> <range rdf:resource="#Female"/> <domain rdf:resource="#Human"/> <label xml:lang="en">has for mother</label> <comment xml:lang="en">to have for parent a female. </comment> </rdf:Property> <hasMother> a rdf:Property ; subPropertyOf <hasParent> ; range <Female> ; domain <Human> .

example of RDF using this schema <rdf:RDF xmlns:rdf ="http://www.w3.org/1999/02/22-rdf- syntax-ns#" xmlns:rdfs="http://www.w3.org/2000/01/rdf-schema#" xmlns="http://inria.fr/2005/humans.rdfs#" xml:base=" http://inria.fr/2005/humans.rdfs-instances" > <rdf:Description rdf:ID="Lucas"> <rdf:type rdf:resource="http://inria.fr/2005/humans.rdfs#Man"/> <hasMother rdf:resource="#Laura"/> </rdf:Description> <Man rdf:ID="Lucas"> <hasMother rdf:resource="#Laura"/> </Man> <Luca> a Man; hasMother <Laura> .

rdfs:label a resource may have one or more labels in one or more natural language <rdf:Property rdf:ID='name'> <rdfs:domain rdf:resource='Person'/> <rdfs:range rdf:resource='&rdfs;Literal'/> <rdfs:label xml:lang='fr'>nom</rdfs:label> <rdfs:label xml:lang='fr'>nom de famille</rdfs:label> <rdfs:label xml:lang='en'>name</rdfs:label> </rdf:Property> <name> a rdf:Property ; range rdfs:Literal ; domain <Person> ; label "nom"@fr, "nom de famille"@fr, "name"@en .

rdfs:comment & rdfs:seeAlso comments provide definitions and explanations in natural language <rdfs:Class rdf:about=‘#Woman’> <rdfs:subClassOf rdf:resource="#Person"/> <rdfs:comment xml:lang=‘fr’>une personne adulte du sexe féminin</rdfs:comment> <rdfs:comment xml:lang=‘en’>a female adult person </rdfs:comment> </rdfs:Class> see also… <rdfs:Class rdf:about=‘#Man’> <rdfs:seeAlso rdf:resource=‘#Woman’/> </rdfs:Class> <Woman> a rdfs:Class ; rdfs:subClassOf <Person> ; rdfs:comment "adult femal person"@en ; rdfs:comment "une adulte de sexe féminin"@fr . <Man> a rdfs:Class ; rdfs:seeAlso <Woman> .

OWLprovides additional primitives for heavyweight ontologies

OWLin one… enumeration intersection union complement  disjunction restriction! cardinality 1..1 algebraic properties equivalence [>18] disjoint union value restrict. disjoint properties qualified cardinality 1..1 ! individual prop. neg chained prop.   keys …

enumerated class define a class by providing all its members <owl:Class rdf:id="EyeColor"> <owl:oneOf rdf:parseType="Collection"> <owl:Thing rdf:ID="Blue"/> <owl:Thing rdf:ID="Green"/> <owl:Thing rdf:ID="Brown"/> <owl:Thing rdf:ID="Black"/> </owl:oneOf> </owl:Class> {a,b,c,d,e}

classes defined by union of other classes <owl:Class> <owl:unionOf rdf:parseType="Collection"> <owl:Class rdf:about="#Person"/> <owl:Class rdf:about="#Group"/> </owl:unionOf> </owl:Class>

classes defined by intersection of other classes <owl:Class rdf:ID="Man"> <owl:intersectionOf rdf:parseType="Collection"> <owl:Class rdf:about="#Male"/> <owl:Class rdf:about="#Person"/> </owl:intersectionOf> </owl:Class>

complement and disjunction complement class <owl:Class rdf:ID="Male"> <owl:complementOf rdf:resource="#Female"/> </owl:Class> declare a disjunction <owl:Class rdf:ID="Square"> <owl:disjointWith rdf:resource="#Round"/> </owl:Class> 

restriction on all values <owl:Class rdf:ID="Herbivore"> <subClassOf rdf:resource="#Animal"/> <subClassOf> <owl:Restriction> <owl:onProperty rdf:resource="#eats" /> <owl:allValuesFrom rdf:resource="#Plant" /> </owl:Restriction> </subClassOf> </owl:Class> !

restriction on some values <owl:Class rdf:ID="Sportive"> <owl:equivalentClass> <owl:Restriction> <owl:onProperty rdf:resource="#hobby" /> <owl:someValuesFrom rdf:resource="#Sport" /> </owl:Restriction> </owl:equivalentClass> </owl:Class> !

restriction to an exact value <owl:Class rdf:ID="Bike"> <subClassOf> <owl:Restriction> <owl:onProperty rdf:resource="#nbWheels" /> <owl:hasValue>2</owl:hasValue> </owl:Restriction> </subClassOf> </owl:Class> !

restriction on cardinality how many times a property is used for a same subject but with different values • Constraints: minimum, maximum, exact number • Exemple <owl:Class rdf:ID="Person"> <subClassOf> <owl:Restriction> <owl:onProperty rdf:resource="#name" /> <owl:maxCardinality>1</owl:maxCardinality> </owl:Restriction> </subClassOf> </owl:Class> 1..1

types of properties • ObjectProperty are relations between resources only e.g. hasParent(#thomas,#stephan) • DatatypeProperty have a literal value possibly typed ex:hasAge(#thomas,16^^xsd:int) • AnnotationProperty are ignored in inferences and used for documentation and extensions

algebraic properties • Symmetric property, xRy  yRx <owl:SymmetricProperty rdf:ID="hasSpouse" /> • Inverse property, xR1y  yR2x <rdf:Property rdf:ID="hasChild"> <owl:inverseOf rdf:resource="#hasParent"/> </rdf:Property> • Transitive property, xRy & yRz  xRz <owl:TransitiveProperty rdf:ID="hasAncestor" /> • Functional property, xRy & xRz  y=z <owl:FunctionalProperty rdf:ID="hasMother" /> • Inverse functional property, xRy & zRy  x=z <owl:InverseFunctionalProperty rdf:ID="hasSocialSecurityNumber" /> ! !

equivalencies and alignment • equivalent classes : owl:equivalentClass • equivalent properties: owl:equivalentProperty • identical or different resources: owl:sameAs, owl:differentFrom 

document the schemas description of the ontology owl:Ontology, owl:imports, owl:versionInfo, owl:priorVersion, owl:backwardCompatibleWith, owl:incompatibleWith versions of classes and properties owl:DeprecatedClass, owl:DeprecatedProperty

OWL profiles EL: large numbers of properties and/or classes and polynomial time. QL: large volumes of instance data, and conjunctive query answering using conventional relational database in LOGSPACE RL: scalable reasoning without sacrificing too much expressive power using rule-based reasoning in polynomial time

VoCamp camps for vocabulary hackers

semantic waste separation the web is a garbage can, the semantic web will be a semantic garbage can.

Rule Interchange Format (RIF) core and extensions

e.g. infer new relations rule: if a member of a team is interested in a topic then the team as a whole is interested in that topic ?person interestedBy ?topic ?person member ?team  ?team interestedBy ?topic interestedByPerson ?person Topic ?topic member Team ?team interestedBy

question: forward chaining ex:Fabien ex:activity ex:Research ex:Fabien ex:in ex:WimmicsTeam ex:WimmicsTeam ex:in ex:INRIASophia ex:INRIASophia ex:in ex:INRIA ex:WimmicsTeam ex:activity ex:Research ex:INRIASophia ex:activity ex:Research ex:INRIA ex:activity ex:Research IF ?x ex:activity ?y ?x ex:in ?z THEN ?z ex:activity ?y

RIF Core subset shared by most systems: add only employee1 [function-> “executive” bonus -> 10 ] ForAll ?emp (?emp [ bonus -> 15 ] :- ?emp [ function -> “executive” ] ) employee1 [function -> “executive” bonus -> 10 bonus -> 15 ]

RIF Core monotonic Horn clause on frames conclusion :- hyp1 and hyp2 and hyp3 … • IRI as constants • frames as triplets • lists • existential quantification in condition • class membership and equality in condition

RIF BLD (Basic Logic Dialect) still monotonic : no changes. • conjunction in conclusion • fonctions, predicates and named arguments f(?x) Maganer(?e) :- Exists ?g (manage(?e ?g)) • disjunction in condition • equality in conclusion • sub-classes

RIF PRD (Production Rules Dialect) full production rules in forward chaining • add, delete, modify, run • instantiate frames (new) • negation as failure (ineg) • no longer monotonic Forall ?customer ?purchasesYTD (If And( ?customer#ex:Customer ?customer[ex:purchasesYTD->?purchasesYTD] External(pred:numeric-greater-than(?purchasesYTD 5000)) ) Then Do( Modify(?customer[ex:status->"Gold"]) ) ) (from PRD Rec. Doc.)

RIF, RIF, RIF,… • DTB (Datatypes and Built-Ins) : data types with their predicates and functions • FLD: how to specify new dialects extending BLD • SWC : syntax and semantics to combine RIF, RDF graphs, RDFS and OWL (RL)

SKOS knowledge thesauri, classifications, subjects, taxonomies, folksonomies, ... controlled vocabulary 168

natural language expressions to refer to concepts 169 inria:CorporateSemanticWeb skos:prefLabel "corporate semantic web"@en; skos:prefLabel "web sémantique d'entreprise"@fr; skos:altLabel "corporate SW"@en; skos:altLabel "CSW"@en; skos:hiddenLabel "web semantique d'entreprise"@fr. labels

between conceptsinria:CorporateSemanticWeb skos:broader w3c:SemanticWeb; skos:narrower inria:CorporateSemanticWiki; skos:related inria:KnowledgeManagement. relations

inria:CorporateSemanticWeb skos:scopeNote "only within KM community"; skos:definition "a semantic web on an intranet"; skos:example "Nokia's internal use of RDF gateway"; skos:historyNote "semantic intranet until 2006"; skos:editorialNote "keep wikipedia def. uptodate"; skos:changeNote "acronym added by fabien".

toward all forms of data on the web

many databuried and dormant in web pages

R2RML a standard transformation of a relationnal database in RDF schema mapping

direct mapping • cells of a line  triples with a shared subject • names of columns  names of properties • each value of a cell  one object • links between tables name fname age doe john 34 did sandy 45 #s1 :name "doe" #s1 :fname "john" #s1 :age "34" #s2 :name "did" #s2 :fname "sandy" #s2 :age "45" #s3 …

example of mapping ISBN Author Title Year 0006511409X id_xyz The Glass Palace 2000 ID Name Homepage id_xyz Ghosh, Amitav http://www.amitavghosh.com http://…isbn/000651409X Ghosh, Amitav http://www.amitavghosh.com The Glass Palace 2000 a:name a:homepage a:author

(1) transforming table of persons ISBN Author Title Year 0006511409X id_xyz The Glass Palace 2000 ID Name Homepage id_xyz Ghosh, Amitav http://www.amitavghosh.com http://…isbn/000651409X Ghosh, Amitav http://www.amitavghosh.com The Glass Palace 2000 a:name a:homepage a:author :P_Table rdf:type rr:TriplesMap ; rr:subjectMap [ rr:termtype "BlankNode" ; rr:column "ID" ; ] ; rr:predicateObjectMap [ rr:predicateMap [ rr:predicate a:name ]; rr:objectMap [ rr:column "Name" ] ] ; rr:predicateObjectMap [ rr:predicateMap [ rr:predicate a:homepage ]; rr:objectMap [ rr:column "Homepage" ; rr:termtype "IRI" ] ] ;

(2) transforming table of books ISBN Author Title Year 0006511409X id_xyz The Glass Palace 2000 ID Name Homepage id_xyz Ghosh, Amitav http://www.amitavghosh.com http://…isbn/000651409X Ghosh, Amitav http://www.amitavghosh.com The Glass Palace 2000 a:name a:homepage a:author :B_Table rdf:type rr:TriplesMap ; rr:subjectMap [ rr:template "http://...isbn/{ISBN}"; ]; rr:predicateObjectMap [ rr:predicateMap [ rr:predicate a:title ]; rr:objectMap [ rr:column "Title" ] ] ; rr:predicateObjectMap [ rr:predicateMap [ rr:predicate a:year ]; rr:objectMap [ rr:column "Year" ; ] ] ;

(3) linking tables ISBN Author Title Year 0006511409X id_xyz The Glass Palace 2000 ID Name Homepage id_xyz Ghosh, Amitav http://www.amitavghosh.com http://…isbn/000651409X Ghosh, Amitav http://www.amitavghosh.com The Glass Palace 2000 a:name a:homepage a:author :B_Table a rr:TriplesMap ; ... rr:refPredicateObjectMap [ rr:refPredicateMap [ rr:predicate a:author ]; rr:refObjectMap [ rr:parentTriplesMap :P_Table ; rr:joinCondition "{child}.Author = {parent}.ID" ] ] ].

schema.org schemas to improve index, search and display e.g: • Creative works, Book, Movie, MusicRecording, Recipe, TVSeries ... • Embedded non-text objects, AudioObject, ImageObject, VideoObject • Event • Organization • Person • Place, LocalBusiness, Restaurant ... • Product, Offer, AggregateOffer • Review, AggregateRating = + + +

RDFa 1.1: example on schema.org <div vocab="http://schema.org/" typeof="Product"> <img rel="image" src="dell-30in-lcd.jpg" /> <span property="name">Dell UltraSharp 30" LCD Monitor</span> <div rel="hasAggregateRating" > <div typeof="http://schema.org/AggregateRating"> <span property="ratingValue">87</span> out of <span property="bestRating">100</span> based on <span property="ratingCount">24</span> user ratings </div> </div> <div rel="offers" > <div typeof="http://schema.org/AggregateOffer"> <span property="lowPrice">$1250</span> to <span property="highPrice">$1495</span> from <span property="offerCount">8</span> sellers </div> </div> (…) PS: RDFa Lite = vocab + typeof + property + about + prefix.

GRDDL opens formats by allowing us to declare RDF extraction algorithms inside XML documents <head profile="http://www.w3.org/2003/g/data-view"> <title>The man who mistook his wife for a hat</title> <link rel="transformation" href="http://www.w3.org/2000/06/ dc-extract/dc-extract.xsl" /> <meta name="DC.Subject" content="clinical tales" /> …

code inside the page <html xmlns="http://www.w3.org/1999/xhtml" dir="ltr" lang="en-US" xmlns:fb="https://www.facebook.com/2008/fbml"> <head prefix="og: http://ogp.me/ns# fb: http://ogp.me/ns# YOUR_NAMESPACE: http://ogp.me/ns/apps/YOUR_NAMESPACE#"> <meta property="fb:app_id" content="YOUR_APP_ID" /> <meta property="og:type" content="YOUR_NAMESPACE:recipe" /> <meta property="og:title" content="Stuffed Cookies" /> <meta property="og:image" content="http://example.com/cookie.jpg" /> <meta property="og:description" content="The Turducken of Cookies" /> <meta property="og:url" content="http://example.com/cookie.html"> <script type="text/javascript"> function postCook() { FB.api('/me/YOUR_NAMESPACE:cook' + '?recipe=http://example.com/cookie.html','post', (…) }); } </script> </head> <body> (…) <form> <input type="button" value="Cook" onclick="postCook()" /> </form> </body> </html>

VoID: describing RDF datasets/linksets

:DBpedia a void:Dataset; void:sparqlEndpoint <http://dbpedia.org/sparql>; void:feature :RDFXML ; void:subset :DBpedia2Geonames ; void:uriLookupEndpoint <http://lookup.dbpedia.org/api/search.asmx/KeywordSearch? QueryString=> ; dcterms:modified "2008-11-17"^^xsd:date; dcterms:title "DBPedia"; dcterms:description "RDF data extracted from Wikipedia"; dcterms:publisher :DBpedia_community; dcterms:license <http://creativecommons.org/licenses/by-sa/3.0/>; dcterms:source <http://dbpedia.org/resource/Wikipedia>. :Geonames a void:Dataset; void:sparqlEndpoint <http://geosparql.appspot.com/query>; void:triples "107983838"^^xsd:integer ; dcterms:subject <http://dbpedia.org/resource/Location> . :DBpedia2Geonames a void:Linkset ; void:linkPredicate owl:sameAs ; void:target :DBpedia ; void:target :Geonames . e.g. DBpedia dataset

Data Cube: publish multi-dimensional data (statistics)

CSV-LD & Linked CSV • contexts to interpret and generate CSV • conventions for CSV to be linked in RDF

SAWSDLsemantic annotation of WSDL (W3C Rec. 2007)

semantically services annotated and searched providerserviceclientrequester directory 3 12

a (too) fast three-tier summary RDFa, microdata,… LDP, HTTP, JSON-LD, … R2RML, SPARQL, RDF, … presentation logic data

Provenance: PROV-DM & PROV-O describe entities and activities involved in providing a resource

PROV-DM & PROV-O: primer example ex:compose prov:used ex:dataSet1 ; prov:used ex:regionList . ex:composition prov:wasGeneratedBy ex:compose . ex:illustrate prov:used ex:composition . ex:chart1 prov:wasGeneratedBy ex:illustrate .

annotating multimédia elements • semantic description of multimedia resources [Media Annotation] • pointing to internal elements of multimedia resources [Media Fragment]

multimedia fragment • part of the URL after the # http://www.example.com/example.ogv#track=audio&t=10,20 • dimensions: – temporal: t=10,20 / t=npt:,0:02:01.5 / t=clock:2009-07-26T11:19:01Z – spatial: xywh=pixel:160,120,320,240 / xywh=percent:25,25,50,50 – track: track=1 / track=video&track=subtitle / track=Wide – named: id=chapter-1 • fragment are not sent with the URL but encoded in the HTTP request

ontologies for multimedia descriptions ontology for Media Resources 1.0 <video.ogv> a ma:MediaResource ; ma:hasTrack <video.ogv#track=audio>, <video.ogv#track=subtitle>; ma:hasSubtitling <video.ogv#track=subtitle> ; ma:hasSigning <video.ogv#xywh=percent:70,70,90,90> . <video.ogv#track=audio> a ma:AudioTrack ; ma:hasLanguage [ rdfs:label "en-GB" ] ; ma:hasFragment <video.ogv#track=audio&t=10,20> . <video.ogv#track=audio&t=10,20> a ma:MediaFragment ; ma:hasLanguage [ rdfs:label "fr" ] . <video.ogv#track=subtitle> a ma:DataTrack ; ma:hasLanguage [ rdfs:label "es" ] . <video.ogv#xywh=percent:70,70,90,90> a ma:MediaFragment ; ma:hasLanguage [ rdfs:label "bfi" ] .

some pointers• W3C standards http://www.w3.org/standards/semanticweb/ • SW Tools http://www.w3.org/2001/sw/wiki/Tools • Linked Data Book http://linkeddatabook.com/editions/1.0/ • W3DevCampus http://www.w3devcampus.com/ • EUCLID material http://www.euclid-project.eu/

http://www.w3.org/2001/sw/wiki/Tools

impossible to predict every usage

explicit make conceptualizations

open your data to those who could use them

#WatchDogs #WeAreData @ubisoft

66 FOAF primitives 3 475 908 348 references (2) x 52 millions “a small tree ruling a big graph”(1) (1) Franck Van Harmelen, ISWC 2011 (2) Libby Miller, 2009

“semantic web” and not “semanticweb” [C. Welty, ISWC 2007] “a lightweight ontology allows us to do lightweight reasoning” [J. Hendler, ISWC 2007]

data data bases data models open data linked data closed data enterprise data linked enterprise data linked open data data schemas semantic web of data data structures linked data schemas web of data big data big data streams data streams linked data streams web of sensors, things, … VELOCITY big linked data VOLUME VARIETY VVeb data linked healthcare data VICINITY VISIBILITY personal data data mining data type

price convert? person homepage? more info? web 1, 2, 3

identify describe & link query reasoning trace URI RDF HTTP, SPARQL, LDP RDFS & OWL PROV-O GOALS AND MEANS

identify describe & link query reasoning trace http://fabien.fr#me #me type man select * {?r type ?t} man subClassOf male wasAttributedTo #me GOALS AND MEANS

informal formal usage representation one web… data person document program metadata

he who controls metadata, controls the web and through the world-wide web many things in our world. fabien, gandon, @fabien_gandon, http://fabien.info WWW 2014

An introduction to Semantic Web and Linked Data

In this document

More Related Content

What's hot

Viewers also liked

Similar to An introduction to Semantic Web and Linked Data

More from Fabien Gandon

Recently uploaded

An introduction to Semantic Web and Linked Data