�ݺ�ߣ

1Franck MICHEL - Université Côte d’Azur, CNRS, Inria, I3S, France
F. Michel
Université Côte d’Azur, CNRS, Inria, I3S, France
Knowledge Engineering:
Semantic web, web of data, linked data
ANF APSEM2018 : Apprentissage et sémantique
Toulouse, 12-15 Nov. 2018

More data sources  More opportunities

To you, your data may mean this…

To others,
your data may mean that…

Interoperability Challenges
Structural heterogeneity
Uniform representation format
Semantic heterogeneity
Controlled vocabularies, thesaurus, ontologies…
Common way to query the data

The Semantic Web
Linked Data and the Web of Data
Publishing legacy data in RDF
Agenda

The Semantic Web

“The Semantic Web provides an environment where
applications can publish and link data, define vocabularies,
query data at web scale, and draw inferences.” (adapted from W3C website)
Link
Querying
Vocabularies
Inference
Publish

Standards of the Semantic Web
Applications and Services
Trust
Identifiers: URI, IRI
Data representation:
RDF abstract model + syntaxes
Vocabularies:
RDFS, OWL, SKOSQuerying:
SPARQL
Rules:
SPIN, SWRL, SHACL
Unifying logic: First Order Logic
Proof
Security(crypto)

Trust
Vocabularies:
SPARQL
Rules:
SPIN, SWRL, SHACL
Proof
Security(crypto)
Web of Data

Trust
Vocabularies:
SPARQL
Rules:
SPIN, SWRL, SHACL
Proof
Security(crypto)
Reasonning

RDF is a conceptual model based on triples,
i.e. any fact consists of 3 components:
( subject, predicate, object )
Source: C. Faron Zucker, O. Corby. Introduction au web de données et au web sémantique. Séminaire INRA Open Data Dec. 2014.
The Resource Description Framework

websem.html is a texte
websem.html has as author Fabien
websem.html has as author Olivier
websem.html has as author Catherine
websem.html has as subject Semantic Web
websem.html was written in 2011

websem.html
SemanticWeb
Texte
Catherine
Olivier
Fabien
type
date
author
subject
author
author
2011

http://ns.inria.fr/
ex/websem.html
http://en.wikipedia.org/
wiki/Semantic_Web
dt:Text
http://ns.inria.fr/
catherine.faron
http://ns.inria.fr/
olivier.corby
http://ns.inria.fr/
fabien.gandon
rdf:type
dc:date
dc:author
dc:subject
dc:author
dc:author
2011

N-Triples
<http://inria.fr/ex/websem.html>
<http://purl.org/dc/elements/1.1/author>
<http://ns.inria.fr/catherine.faron> .
<http://purl.org/dc/elements/1.1/theme> "Semantic Web" .
@prefix dc: <http://purl.org/dc/elements/1.1/> .
dc:author <http://ns.inria.fr/catherine.faron> ;
dc:theme "Semantic Web" .
Turtle
RDF Syntaxes: N-Triples, Turtle, JSON-LD, Trig, RDF/XML…

RDF Schemas define
classes of resources,
their properties,
and organize their hierarchies

igeo:TerritoireAdministratif
igeo:Commune
rdfs:subClassOf rdfs:Class
rdf:type
rdf:type
http://id.insee.fr/geo/
commune/34172
rdf:type
@prefix igeo: <http://rdf.insee.fr/def/geo#> .
RDF Schema - Classes

igeo:codeINSEE
igeo:codeCommune
rdfs:subPropertyOf rdf:Property
rdf:type
rdf:type
RDF Schema - Properties

igeo:Commune
rdfs:range
igeo:chefLieu
igeo:PaysOuTerritoire
rdfs:domain
departement/34
igeo:chefLieu
rdf:typerdf:type
commune/34172
Montpellier
RDF Schema - Properties

OWL
The Web Ontology Language

def. by enumeration
def. by intersection
def. by union
def. by complement
 class disjunction
def. by restriction
def. by cardinality
def. by equivalence
!
1..1

[>=18] def. by value restrict.
…
OWL in one slide…
(a)symetric prop.
prop. disjunction
cardinality1..1
!
indiv. prop. negation
chained prop.


(irr)reflexive prop.
transitive prop.
inverse prop.

Closed vs. Open Worlds Assumptions
Closed World
Everything there is to know about a thing is
stated in a single, closed DB.
 Not asserted facts are false, i.e.
only asserted facts are true.
 A schema may define what can be stated
(a schema may be violated).
Open World
Knowledge is distributed.
Each RDF graph may state facts about a thing,
irrespective of what others state.
 Because a fact is not asserted does not
mean it is false.
 Every asserted fact is true (no schema)
 But some facts may lead to inconsistencies

Quering RDF with SPARQL
SPARQL Protocol and RDF
Query Language

SPARQL 1.1 Rec. 21 Mar. 2013
 Query Language (using the Turtle syntax)
• CRUD operations
 Query results
• Query Results Format XML, JSON, CSV/TCV
 Protocols
• SPARQL Protocol
• SPARQL Graph Store HTTP Protocol
 Entailment Regimes

SPARQL: triple patterns
Turtle syntax with “?” or “$” to mark variables:
?x rdf:type ex:Person
Describe patterns of triples that we look for:
SELECT ?subject ?type
WHERE { ?subject rdf:type ?type }
Default pattern: conjunction of triple patterns:
SELECT ?x WHERE
{ ?x rdf:type ex:Person .
?x ex:name ?name . }
?x
rdf:type
ex:Person
?name
ex:name

SPARQL: namespace prefixes
Declare prefixes of used vocabularies:
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> .
}
Declare a base namespace for relative URIs:
BASE <http://example.org/people#>
SELECT ?student
WHERE { ?student foaf:knows <Ted> . }
?student
mit:registeredAt
?x
http://www.mit.edu
foaf:homepage
http://example.org/
people#Ted
foaf:knows

SPARQL: language and typed literals
SELECT ?x ?f WHERE {
?x foaf:name "Steve"@en ; foaf:knows ?f .
}
SELECT ?x WHERE {
?x foaf:name "Steve"@en ;
foaf:age "21"^^xsd:integer .
}

SPARQL: optional pattern
SELECT ?person ?name
WHERE {
?person foaf:homepage <http://fabien.info> .
OPTIONAL { ?person foaf:name ?name . }
}
 Variable ?name is potentially unbound.

SPARQL alternative pattern
Merge the results of two graph patterns:
WHERE {
?person foaf:name ?name .
{ ?person foaf:homepage <http://fabien.info> . }
UNION
{ ?person foaf:homepage <http://fabien.org> . }
}

SPARQL filters
PREFIX ex: <http://inria.fr/schema#>
WHERE {
?person rdf:type ex:Person; ex:name ?name; ex:age ?age .
FILTER (xsd:integer(?age) >= 18)
}
Other examples:
FILTER(?name IN ("fabien", "olivier", "catherine"))
FILTER(langMatches(lang(?name),"en"))

SPARQL additional features
 Solution modifiers:
ORDER BY, LIMIT, OFFSET, DISTINCT
 Aggregates
GROUP BY, HAVING
 Negation
NOT EXISTS, MINUS, NOT IN
WHERE { ?x a ex:Person MINUS { ?x foaf:knows ex:John } }
 Nested queries
 Named graphs
 Property paths
?x foaf:knows+ ?friend .
 …

SPARQL JSON results
{
"head": { "vars": [ "student" ] },
"results": {
"bindings: [
{"student": {
"type": "uri",
"value": "http//www.mit.edu/data.rdf#joe" }
},
{ "student": {
"type": "uri",
"value": "http//www.mit.edu/abcdef" }
}
]
}
}

PREFIX igeo:<http://rdf.insee.fr/def/geo#>
SELECT ?x
WHERE { ?x rdf:type igeo:TerritoireAdministratif }
igeo:TerritoireAdministratif
igeo:Commune
rdfs:subClassOf
ex:Montpellier
rdf:type
SPARQL Entailments: infer knowledge

PREFIX igeo:<http://rdf.insee.fr/def/geo#>
SELECT ?x ?code
WHERE { ?x igeo:codeINSEE ?code}
igeo:codeINSEE
igeo:codeCommune
rdfs:subPropertyOf

SELECT ?x WHERE { ?x rdf:type igeo:Commune }
SELECT ?x WHERE { ?x rdf:type igeo:PaysOuTerritoire }
igeo:Commune
rdfs:range
igeo:chefLieu
igeo:PaysOuTerritoire
rdfs:domain
departement/34
igeo:chefLieu http://id.insee.fr/geo/
commune/34172
rdf:typerdf:type

The Semantic Web
Agenda

The Web of Data
aka. Data Web, Web 3.0,
Global Knowledge Graph…

The Web of Data
Trust
Vocabularies:
SPARQL
Rules:
SPIN, SWRL, SHACL
Proof
Security(crypto)
First step in the deployment
of the Semantic Web
Detractors would say
the part of the
Semantic Web that works…

“The Semantic Web provides an environment where
applications can publish and link data, define vocabularies,
query data at web scale, and draw inferences.”
Link
Querying
Vocabularies
Inference
Publish

Linked Data principles
1.Use URIs to name things
2.Use HTTP URIs so that people
can look up those names
3.When someone looks up a URI,
provide useful information using the standards (RDF, SPARQL)
4.Include links to other URIs, so they can discover more things

Linked Open Data Cloud: 1200+ linked datasets
Linking Open Data cloud diagram, 2018. J.P. McCrae, A. Abele,
P. Buitelaar, A. Jentzsch, V. Andryushechkin and R. Cyganiak.
http://lod-cloud.net/
 On the web, under open licenses
 Machine-readable (RDF)
 URIs to name things
 Common vocabularies
 Linked with each other
 Queryable
Iconic but partial view of the Web of Data
LOD Atlas: 25,000+ datasets

The Semantic Web
Agenda

Publishing legacy data in RDF raises tricky questions
Legacy
dataset

Metadata
Data
Legacy
dataset
describe

Metadata
Data
Legacy
dataset
describe
Catalogue,
data portal
What metadata?
Where/how to publish them?

Metadata
Data
Ensure shared
understanding?
Legacy
dataset
describe
Catalogue,
data portal
What metadata?

Metadata
Data
Ensure shared
understanding?
Reference raw data
(signals, binary…)
Legacy
dataset
describe
Catalogue,
data portal
What metadata?

Metadata
Data
Ensure shared
understanding?
Reference raw data
Translate
heterogeneous
data into RDF?
Legacy
dataset
describe
Catalogue,
data portal
What metadata?

Ensure shared understanding?
Need for common vocabularies with well defined semantics
• Controlled vocabulary, thesaurus, ontology
• How to define/model a vocabulary?
• Where to find existing vocabularies, how to reuse and/or them?

Metadata
Data
Ensure shared
understanding?
Reference raw data
Translate
heterogeneous
data into RDF?
Legacy
dataset
describe
Catalogue,
data portal
What metadata?

Many methods for many types of data sources
AstroGrid-D, SPARQL2XQuery, XSPARQL
XML
XLWrap, Linked CSV, CSVW, RML
CSV/TSV/Spreadsheets
D2RQ, R2O, Ultrawrap, Triplify, SM
R2RML: Morph-RDB, ontop, Virtuoso
Relational Databases
RML, TARQL, Apache Any23, DataLift,
SPARQL-Generate
Multiple formats
RDFa, Microformats
HTML
TARQL, JSON-LD, RML
JSON
xR2RML (MongoDB), ontop (MongoDB),
[Mugnier et al, 2016] (key-value stores)
NoSQL
M.L. Mugnier, M.C. Rousset, and F. Ulliana. “Ontology-Mediated Queries for NOSQL Databases.” In Proc. AAAI. 2016.
SPARQL Micro-services, Linked REST APIs
Web APIs

Metadata
Data
Ensure shared
understanding?
Reference raw data
Translate
heterogeneous
data into RDF?
Legacy
dataset
describe
Catalogue,
data portal
What metadata?

 Metadata vocabularies
Schema.org, DCAT, VOID, HCLS
 Data portals and catalogues
CKAN, data.gov.*, Google Dataset Search
Vocabularies to describe datasets and dataset catalogues

Thank you!

�ݺ�ߣ

Knowledge Engineering: Semantic web, web of data, linked data

More Related Content

Knowledge Engineering: Semantic web, web of data, linked data