Upload
minel-jean-luc
View
466
Download
0
Embed Size (px)
DESCRIPTION
we adopt two points of view. 1) A national contributor point of view who wants to give visibility to his contributions 2) A researchers point of view, of any countries, who is looking for specific tools or data of any topics. Consequently, this proposal wants to give to the researchers some means to find relevant information
Citation preview
VCC3Proposal
Displaying and FindingJean-Luc Minel (MoDyCo, Univ. Paris Ouest-La Défense & TGE Adonis)In collaboration with Sophie David, Shadia Kilouchi, Nicolas Larrousse,
Stéphane Pouyllau (TGE Adonis) and Laurent Capelli (CCSD)
22-23 May 2013
“Improve research opportunities and outcomes through linking distributed digital source materials of many kinds”
http://www.dariah.eu/
� For contributorsTo give visibility to their contributions
� For researchersTo give them tools to find relevant information
Objectives
2
� Who are experts on Open Archive?� Who offers a PID service?� Who works on Alexandrian pottery, 2nd century B.C.?� What are the available collections on archeology?� What is the procedure to obtain the DSA?� What are the recommended formats for images? � What are the Dutch contributions?� Is Jean-Luc Minel involved in Dariah?� Is the INA (Institut national de l’audiovisuel) involved in
Dariah?� Which European projects are related to Dariah?� etc.
What could be relevant questions?
3
� To deal with decentralized dataEach contributor is responsible for the description of his contribution
Each country is responsible for gathering and displaying the contributions
� To use standard toolsTo use languages of the Semantic Web (RDF, SPARQL)
� To exploit Linked Open Data possibilitiesTo use existing data from other repositories
� Low cost and time investment
Principles
4
Workflow
5
Proof of concept
6
Some details
Example of RDFa Annotations
<!-- la description du contenu de la contribution --><meta property="dc:subject" content="type d'offre : Accès" /><meta property="dc:subject" content="DARIAH" /><meta property="dc:subject" content="Linguistique" /><meta property="dc:subject" content="Histoire" />
<meta property="dc:subject" content="VCC3" /><meta property="dc:subject" content="Corpus journalistique, Presse Régionale, PQR, XML - TEI P5, TEI P5, Est Républicain, Productivité" />
7Name of the VCC Type of offerDiscipline Discipline
Some rough details
SPARQL Query
PREFIX rdfs: <http://www.w3.org/2000/01/rdf-schema#> PREFIX foaf: <http://xmlns.com/foaf/0.1/> PREFIX dcterms: <http://purl.org/dc/terms/> PREFIX dc: <http://purl.org/dc/elements/1.1/> PREFIX skos: <http://www.w3.org/2004/02/skos/core#>
select DISTINCT ?title ?name ?subject {?x rdfs:type <http://www.rechercheisidore.fr/class/Source> .?x skos:altLabel "DARIAH FR" .?uri ?p ?x. ?uri dcterms:title ?title .?uri dcterms:creator ?contact .?contact foaf:name ?name .?uri dc:subject ?subject .?uri dc:subject "VCC3"@fr.
}
8
Answering relevant questions with a smarter HCI
SPARQL Query http://sandbox-ist.tge-adonis.fr/dariah-fr/demo/
9
Answering relevant questions
PID assigned by ISIDORE if necessary
Automatic enrichments using thesaurus and skos relations10
A very smart HCI
(http://www.rechercheisidore.fr/search/?source=10670/2.8uuv54)
11
Example with VIAFhttp://www.oclc.org/content/dam/research/presentations/hickey/20110302-EMEARC.pdf
To benefit from Linked Data
12
To benefit from Linked Data
13
To benefit from Linked Data
14
To benefit from Linked Data
15
geonames.org
creativecommons.org
dbpedia.orglexvo.org 16
<meta property=‘dc:coverage’ content=‘http://dbpedia.org/page/France’ />
17
Some milestones
� How long to make annotations using RDFA ?
� Between 15 or 30 mn by contributions (depending on who make it and the accuracy of the metadata)
� How long to develop a crawler ?
� No need to develop a crawler. ISIDORE exists and is available (French contribution in Dariah). Of course, it is possible to use another crawler.
� How long to build a triplestore?
� Few hours using a private or public data center. It is not required that each country builds a Tstore.
� How long to develop simple HCI ?
� One day by an agile digital humanist. Of course, HCI can be share18
Flexibility and Responsibility/Best practices
� Dariah.eu can display all contributions on its website
AND
� All partners can display and expand all their contributions with their own choices (VIAF, IDREF, Geonames, Pactols, etc.) and with their own interfaces
***
� As all partners describe and expand their contributions, they are responsible for their visibility... which is also a best practice
19
Some issues
� Contributions in English
� “Standardisation” of the description of the contributions (proposition of a template)
� Choice of vocabularies
Dcterms, foaf, skos, bibo
� Taxonomies, ontologies and thesauri
Ex.: NeDiMAH ontology, Rameau, Geonames, etc.
Existing, simple and but not perfect!
20
Some issues
Linked DataHow to exploit data from other triplestores?Which ones ?
Social Network?
21
In a nutshell
� Each partner manages its contributions and displays them on a webpage of a website� Each webpage is annotated with RDFa, following some
guidelines (using common tags and vocabularies)
� Dariah.eu (and/or Dariah.Anycountry) harvests these websites regularly and puts all the harvested data in a triplestore
� Dariah.eu and/or Dariah.Anycountry offer simple tools to peruse all these data
� Anyone can search in the triplestore using Sparql queries
� Visibility, simplicity, interoperability
22
Huma-NumVery Large Facility for the Digital Humanities
Produced by