Unlocking the full potential of five-star addresses by using Linked Data Fragments
Transcript of Unlocking the full potential of five-star addresses by using Linked Data Fragments
OVERCOMING THE CATCH-22 DILEMMAUNLOCKING THE FULL POTENTIAL OF LOCATION-BASED SERVICES AS LINKED OPEN DATA IN REAL-TIME ECOSYSTEMS.
Raf Buyle, Ziggy Vanlishout Cristian Vasquez Paulus and Pieter ColpaertINSPIRE – 2017 - Strasbourg France
Flanders Information Agency
The cost for re-using public sector information is too high.
Scalable services are expensivefor public administrations
The 5-star Flemish Address Registrylessons learned
The Address Base Registry contains well over 4 million addresses and their geographical coordinates.
http://data.vlaanderen.be/id/adres/2584882
Persistent Identifiers
URIs1
Dereferenceable HTTP
URIs
Human readable
Links to other
information1
Machine friendly
5-STAR Linked Open Data The Web as a Blueprint
information in line with
standards (RDF) 1
1 Tim Berners-Lee, https://www.w3.org/DesignIssues/LinkedData.html
Linked Data Address Products
A data dump• all triples in an entire dataset• http://data.vlaanderen.be/dumps
A subject page • triples about a specific subject in a dataset• http://data.vlaanderen.be/id/adres/2584882
A SPARQL endpoint• triples that correspond to a SPARQL query• http://data.vlaanderen.be/sparql
CPU-cost raises linearly with amount of queries
Resources on the web, identified by urls through the uniform interface offered by HTTP can be cached easily
SPARQL endpoints implement a protocol on top of HTTP, therefore regular HTTP caching can not be applied2
2 Verborgh R. et al. http://linkeddatafragments.org/publications/ldow2014.pdf
Verborgh R. et al. http://linkeddatafragments.org
Can we balance the CPU-cost between the client and the server?
A comparison of required processing (filledbars) and datatransfer (dottedlines) shows why ldf scales significantly better.Verborgh R. et al
Addresses LDF• > 4 Mio addresses are published as Linked Data Fragments • RDF documents can be compared to ‘tiles’• Documents contain links to other documents, which allow machines to
browse the dataset• The document-fragments are visualized on a map,
and cached using the standard HTTP infrastructure
TAKE AWAY
@info_vlaanderen @rafke @ziggyvanlishout @pietercolpaert
Scaling Linked Data SPARQL endpoints is expensive
Overcome catch 22 by reassessing client server trade-offs.
Linked Data Fragments, offer a hybrid solution.
New querying paradigm
Demo: https://bit.ly/geo_ldf API: https://bit.ly/geo_ldf_api
Flemish GovernmentLDF initiatives
Geospatial LDF – Address Registry• Demo: https://bit.ly/geo_ldf • API: https://bit.ly/geo_ldf_api
Heritage Thesauri in Flanders - LDF• Online editor: loveable user interface• Open API for reading and writing• Demo: https://thesaurus.onroerenderfgoed.be• Source: https://github.com/OnroerendErfgoed/atramhasis