Linked Open Data Utrecht University Library
- 1. December 2021 dr. Ruben Schalk Subject Specialist History & Digital Humanities Utrecht University Library r.schalk@uu.nl https://www.uu.nl/staff/RSchalk Linked (Open) Data for researchers and libraries: intro & showcase
- 2. dr. Ruben Schalk WHY THIS PRESENTATION Subject Specialist History & Digital Humanities Utrecht University Library r.schalk@uu.nl • (University) Libraries ideally positioned: Metadata experts Disclosure and accessibility of collections Different collections & formats Networks to facilitate use of metadata standards Open Science & Open Access FAIR research support/ workflow Research output management Links up with Digital Humanities support • But requires additional skills
- 3. Today: 1. What is Linked Data? 2. Why should I use it? 3. Demo Time!
- 4. 1. What is Linked Data? ‘Linked Data is structured data which is interlinked with other data so it becomes more useful through semantic queries’ Source: Wikipedia I almost get it…?
- 5. 1. What is Linked Data: example c_code country_name gdp_capita year 634 Qatar 156029 2015 578 Norway 82713 2015 784 United Arab Emirates 74746 2015 414 Kuwait 71354 2015 702 Singapore 65660 2015 756 Switzerland 59307 2015 442 Luxembourg 55972 2015 372 Ireland 54278 2015 840 United States 52591 2015 702 Singapore 65660 2015 ?
- 6. 1. What is Linked Data: from rows to graphs 702 Singapore 65660 2015 Singapore Country 65660 2015 702 rdf:type clio:hasGDP schema: observationDate schema:country_code Data are now semantically defined: Codebook inherent to data Human readable Machine readable “code that represents a country” skos:label prov:wasDerivedFrom DOI to paper about this code
- 7. 1. What is Linked Data: travel linked data graph Singapore Country 65660 2015 702 rdf:type clio:hasGDP schema: observationDate schema:country_code owl:sameAs Wikipedia: Singapore schema:country_code Some other economic indicators on Singapore in another dataset
- 8. 1. What is Linked Data: building blocks Singapore Wikipedia: Singapore owl:sameAs 2015 SUBJECT PREDICATE OBJECT Basically: a statement or a fact schema:observationDate Often written as N-triples: <https://uu.nl/datasets/mydata/country/Singapore> <http://schema.org/observationDate> <“2015”^^xsd:gYear> . <https://uu.nl/datasets/mydata/country/Singapore> <http://www.w3.org/2002/07/owl#sameAs> <https://en.wikipedia.org/wiki/Singapore> Or use prefixes: mydata:Singapore schema:observationDate “2015” ”^^xsd:gYear
- 9. 1. What is Linked Data: building blocks SUBJECT PREDICATE OBJECT Basically: a statement or a fact The elements of a triple are URI references, literals (or blank nodes): • URI references: a standardized way to identify objects (often online): ISBN, URL, DOI, email address, places, landmark, etc. That does not work for things like numbers, that have multiple meanings in different contexts… • Literals: data values such as strings, dates, integers, decimals, etc. Type of literal is specified inside the triple, remember <“2015”^^xsd:gYear> ?
- 10. 1. What is Linked Data? Linked Data graph = Combination of triples = That point to inside (dataset/ collection) and outside information - ideally defined using common standards = Internet as a global database where everything is connected
- 11. 1. What is Linked Data: how to access it • Many linked data services run in the background of websites • Linked data browsers provide facetted browsing over graph patterns • Specify the graph patterns you want to retrieve with SPARQL queries: PREFIX dbo: <http://dbpedia.org/ontology/> SELECT ?capital ?RESULT WHERE { ?RESULT dbo:capital ?capital . ?capital dbo:areaCode "030" . } ORDER BY DESC(?capital) LIMIT 100 subject predicate object
- 13. Can make it as complex as you like Store queries online (e.g. Github) for sharing and replication! Use API calls to present results to end user
- 14. 2. Why should I use it? Researchers: Linked Data = FAIR data! 5-star data: ★ make your stuff available on the Web (whatever format) under an open license ★★ make it available as structured data (e.g., Excel instead of image scan of a table) ★★★ make it available in a non-proprietary open format (e.g., CSV instead of Excel) ★★★★ use URIs to denote things, so that people can point at your stuff ★★★★★ link your data to other data to provide context, and benefit from the network effect Source: https://5stardata.info/en/
- 15. 2. Why should I use it? Researchers: Linked Data = FAIR data! But also enhances your workflow: Annotate, code and harmonize data in one go, using community standards Share dataset, script (SPARQL), and results live on the Web Answer new questions & find novel patterns by interlinking datasets Run analyses across multiple datasets at the same time No need for codebooks or complex relational queries: it’s all in the data! Graph data model suited to heterogeneous or sparse data Replicable research Easy collaboration
- 16. 2. Why should I use it? Libraries: Superior Search & Find: • Execute very detailed searches by combining metadata • Search across different formats: datasets, books, illustrations, maps, archives, music, etc. • Connect different materials: maps to books, journal papers to related datasets and code, etc. • Easy to embed (part of) catalogue on website • Prioritized by Google (schema.org vocabulary) Link items/catalogues to all types of external data, yet keep them separate Contextualize search results, enhance metadata, or recommend stuff Concentrate on your own expertise Use graph patterns to ascertain quality of the metadata Generic tools instead of domain-specific software for cataloguing
- 18. Research: what if we combined datasets on historical stature as Linked Data? • Initiated by Prof. Joerg Baten (University of Tuebingen) • Shows added value of linking various small to large N datasets centering around the same topic • Possibilities with Linked Data: Link to Clio-Infra LOD dataset to get GDP: correlate average height and GDP before 1950; analyzing all 32 datasets, or 380,000 observations at once! Link to C-shapes LOD for maps: average stature around the world visualized. • Available at: https://druid.datalegend.net/dataLegend/microHeights
- 21. Research: use LOD to study excess mortality during the Spanish Flu epidemic? • CSV on deaths 1910-20 converted to Linked Open Data (using COW) • Harmonized using other LOD datasets • GIS added using yet another LOD dataset • Research output published live on the Web • Downloadable results Carpenter Deceased: Carpenter jobhoard:occupation HISCO:95490 jobhoard:HISCO 52,50 jobhoard:HISCAM = indicator for social economic status
- 23. gemeentegeschiedenis: municipality Utrecht geo:polygon (for year of deceased) geo:hasGeometry Deceased mydata:hasLocation Utrecht Connect to municipality dataset for automatic GIS visualization
- 24. Research: Infant mortality in 19th c. Amsterdam • Project on infant mortality (Radboud University, Prof. Angelique Janssens) • Street-level information on births and deaths • Neighborhoods retrieved from Amsterdam Time Machine Simply connect!
- 25. Libraries: find more relevant items using Linked Data
- 27. Author in Worldcat Linked Data Find related works {by literary movement} in DBpedia Take the identifiers of those works to Wikidata Get Worldcat link for every work = Linked Data graph
- 29. Beats the Google Knowledge Panel!
- 30. If you are interested in this: You might also like: {insert titles or authors from our earlier query here} Run SPARQL query in the background and use machine-readable semantic relations for automated suggestions:
- 32. Collections: give me the different types of work associated with Karl Marx, using IISH knowledge graph: SELECT ?type (COUNT(?work) as ?n) WHERE { ?topic a schema:Person . ?topic schema:name ?name . ?work schema:about ?topic . ?work rdf:type ?type . FILTER(REGEX(?name, "Marx, Karl")) } ORDER BY DESC (?n) Connect anything you like type n http://purl.org/dc/dcmitype/Text 2114 http://purl.org/dc/dcmitype/StillImage 747 http://schema.org/Photograph 202 https://iisg.amsterdam/vocab/Poster 189 https://iisg.amsterdam/vocab/Print 142 https://iisg.amsterdam/vocab/Drawing 117 http://purl.org/dc/dcmitype/PhysicalObject 74 http://schema.org/CreativeWork 67 https://iisg.amsterdam/vocab/Postcard 61 http://purl.org/dc/dcmitype/Collection 7 http://purl.org/dc/dcmitype/Sound 7 http://schema.org/Collection 7 https://iisg.amsterdam/vocab/ImageCollection 1 http://schema.org/CreativeWorkSeries 1 http://schema.org/Game 1
- 33. Use URI’s to connect information from different collections Place and/or time : Give me all information on Amsterdam in the year 1790: Source: http://years.amsterdamtimemachine.nl/?year=1790 Special collections: Give me all digitized versions of Blaeu’s Atlas Major across libraries: <http://dbpedia.org/resource/Atlas_Maior> <dbo:wikiPageExternalLink> ?url_to_work . Connect anything you like <https://utrechtuniversity.on.worldcat.org/oclc/901235386> <https://www.erfgoedleiden.nl/schatkamer/bladeren-door-blaeu> <http://digital2.library.ucla.edu/viewItem.do%3Fark=21198/zz0017r9p5> <http://digital.ub.uni-duesseldorf.de/urn/urn:nbn:de:hbz:061:1-37297> <http://maps.nls.uk/atlas/blaeu/>
- 34. • And use URI’s to put your collection in context I have a picture of a railway station. How do I find out who’s the architect if that’s not in the metadata…? • Link station URI with another catalogue, and improve metadata! Source: https://api.data.netwerkdigitaalerfgoed.nl/s/QY9kX9nB https://github.com/RubenSchalk/grlc-test/blob/master/hua_beeldbank_architects.rq Connect anything you like
- 35. Ask any question you like • Royal Dutch library (KB): ‘which animals featured most in novels by Dutch women writers since the 1980s?’ Source: https://data.netwerkdigitaalerfgoed.nl/enno/-/ queries/Dieren-en-vrouwen/4 • DBpedia: ‘soccer players who were born in a country with more than 10 million inhabitants, who played as goalkeeper for a club that has a stadium with more than 30,000 seats, and whose club country is different from their birth country’ animal count cats 314 birds 240 dogs ; separate breeds 188 dogs 180 aviary birds 80 cats ; separate breeds 63 soccerplayer countryOfBirth team countryOfTeam stadiumcapacity Losseny_Doumbia Niger Daring_Club_Motema_Pembe Democratic_Republic_of_Congo 80000 Arakaza_MacArthur Burundi Lusaka_Dynamos_F.C. Zambia 60000 Daniel_Ferreyra Argentina FBC_Melgar Peru 60000 Mohammed_M._Tagoe Ghana Lusaka_Dynamos_F.C. Zambia 60000 Sunday_Rotimi Nigeria Mekelle_70_Enderta_F.C. Ethiopia 60000 Anthony_Scribe France FC_Dinamo_Tbilisi Georgia_(country) 54549 Zaur_Khapov Russia FC_Dinamo_Tbilisi Georgia_(country) 54549 Jose_Carlos_Fernnndez Bolivia Deportivo_Cali Colombia 44000 Leonardo_Daaz Argentina Deportivo_Cali Colombia 44000
- 36. To conclude: some useful links Utrecht University Library: https://www.uu.nl/en/university-library UU Library Digital Humanities Support: https://www.uu.nl/en/university-library/advice- support-to/researchers/digital-humanities-support Royal Dutch Library: https://www.kb.nl/bronnen-zoekwijzers/dataservices-en-apis/linked- data-van-de-kb Netwerk Digitaal Erfgoed: https://netwerkdigitaalerfgoed.nl/activiteiten/linked-data/ Make your own Linked Data: • https://ldwizard.netwerkdigitaalerfgoed.nl/ • https://github.com/CLARIAH/COW • https://marcedit.reeset.net/ SPARQL 101: http://www.learningsparql.com/ Open source SPARQL interface: http://yasgui.triply.cc/ Generate API calls on SPARQL queries: http://grlc.io/