Publishing Ordnance Survey Ireland's geospatial data as Linked Data Prof. Declan O’Sullivan (declan.osullivan@scss.tcd.ie) School of Computer Science and Statistics ADAPT @ Trinity College Dublin, Ireland The ADAPT Centre is funded under the SFI Research Centres Programme (Grant 13/RC/2106) and is co-funded under the European Regional Development Fund.
Geopspatial Linked Data Project co-funded with Ordnance Survey Ireland (OSi) Goal: Develop a semantic architecture and Linked Data platform for the OSi taking into account best practices and guidelines in the domain of geospatial information and industry and OSi’s current technology stack. Started with the boundaries dataset, which was open and already available on data.gov.ie, but not as Linked Data.
OSi’s authoritative geospatial platform Over 50 million uniquely reference features Single source of truth (scale independent) Maintained object lifecycles Real world data model Non-proprietary Based on industry standards
What had to be done? Features and Geometries with GeoSPARQL Modelling Provenance of Statute Instruments with PROV-O Ontologies developed and published Workshops and agreement with DPER and CSO on URI Strategy Uplift the Prime2 data into RDF with declarative R2RML mappings GeoSPARQL Reference http://www.opengeospatial.org/standards/geosparql W3C R2RML Reference https://www.w3.org/TR/r2rml/
Architecture of the LD Platform Linked Data Frontend (Pubby) Proxy Server Application or Web Browser Triple Pattern Fragments Server Triple Pattern Fragments Web Client Triple Pattern Fragments Client RDF /Turtle Data Dumps Ontologies Web Browser SPARQL EndPoint TripleStore (Parliament) R2RML Mappings PRIME 2 (Oracle) 5 STAR LINKED OPEN DATA STATUS ACHIEVED
Uplift from RDB to RDF COUNTY (table in CSV) R2RML Mapping GUID FIRST_CO_ENGLISH FIRST_CO_GAEILGE GEOM 2AE19629144813A3E055000000000001 CAVAN An Cabhán POLYGON ((-7.27981205778366 54.1025770715918 ... )) 2AE19629144E13A3E055000000000001 WICKLOW Cill Mhantáin MULTIPOLYGON (((-6.03209172629314 52.9822024015277 … ))) 2AE19629144713A3E055000000000001 ROSCOMMON Ros Comáin POLYGON ((-8.6703692379984 53.7722704863694 ... )) R2RML Mapping <#TriplesMap1> rr:logicalTable [ rr:tableName "COUNTY" ; ] ; rr:subjectMap [ rr:template "http://data.geohive.ie/resource/county/{GUID}" ; rr:class geohive:County ; ]; rr:predicateObjectMap [ rr:predicate rdfs:label ; rr:objectMap [ rr:column "FIRST_CO_ENGLISH" ; rr:language "en" ;]; ]; rr:objectMap [ rr:column "FIRST_CO_GAEILGE" ; rr:language "ga" ;]; ]; rr:predicateObjectMap [ rr:predicate geo:hasGeometry ; rr:objectMap [ rr:parentTriplesMap <#TriplesMap2> ; rr:joinCondition [ rr:child "GUID" ; rr:parent "GUID" ;] ]; ]. <#TriplesMap2> rr:logicalTable [ rr:tableName "COUNTY" ; ]; rr:subjectMap [ rr:class geo:Geometry ; rr:termType rr:BlankNode ; rr:column "GUID" ; ]; rr:predicateObjectMap [ rr:predicate geo:asWKT ; rr:objectMap [ rr:column "GEOM" ; rr:datatype geo:wktLiteral ;] ]. <http://data.geohive.ie/resource/county/2AE19629144813A3E055000000000001> a geohive:County ; rdfs:label "CAVAN"@en , "An Cabhán"@ga ; geo:hasGeometry [ a geo:Geometry ; geo:asWKT "POLYGON ((-7.27981205778366 54.1025770715918 ... )"^^geo:wktLiteral ] . Generated RDF in Turtle format
Sample generated RDF in Data Dump
Parliament Triple store and SPARQL processor Supports reasoning, but importantly supports GeoSPARQL (OGC standard) Provides functions over geometries (within, touches, distance etc.) Useful for geospatial analyses Parliament Reference http://parliament.semwebcentral.org
data.geohive.ie
Pubby Provides a front end for Linked Data Sits on a SPARQL endpoint RDF browser access data Publish on web for human and machine consumption Flexible and customisable Pubby Reference http://wifo5-03.informatik.uni-mannheim.de/pubby/
Implementation: Linked Data Frontend Description of County Clare linking to its three representations
Implementation: Linked Data Frontend Default generalization with OSi’s base map. Different representations
Triple Pattern Fragments Query Linked Data with less load on server Reduce server side processing load as client does more TPF client and server communicate User writes query on client which sends triple pattern requests to server Server sends back data based on triple pattern to client Client then performs processing intense parts over the returned data Triple Pattern Fragments Reference https://biblio.ugent.be/publication/8050661
Implementation: Triple Pattern Fragment Client
Proof of Concepts
Proof of Concept: Access Control to Closed Linked Data Licenses: data customer can access Templates: SPARQL based template than can be used for access Template Selector: shows allowed templates (based on licenses) Template Analyser: checks template variables against license SPARQL Processor Templates and Licenses (RDF) Access Control Module RESTful API Template Selector Template Analyser Query Processor SPARQL Processor GeoHive (RDF) Related research paper under submission to ESWC 2018 4
Proof of Concept: OSi and Building Information Modelling Building Information Modelling (BIM) is a key enabler for integrating building data across a buildings life cycle (BLC) and with other related data sources energy, occupancy, weather, etc. Ordnance Survey Ireland (OSi) Prime2, includes not only GIS data, but also additional building specific data (form and function). Uplifting this building data will provide an authoritative basis for interlinking data about buildings in Ireland supporting an iterative approach to generating BIM and opening up additional markets for OSi. Research Paper Reference https://iswc2017.semanticweb.org/paper-557/
Integrating OSi Building Data based on Geolocation THE VISION Data Uplift Tabular Data Identify Alignments Generate RDF Data in RDF (OSi Building + CSO Data) Data Conversion IFC STEP Ifc->owl Ifcowl- > ifcowl_geo IFC in RDF with added Geolocation Data Conversion DBPedia RDF SPARQL CONSTRUCT DBpedia with added Geolocation Integrated Building Information OSi-Geometry: Footprint OSi-Form: Building General OSi-Function: Residence LD DBpedia: Info on Architect LD IFC: Structure (walls, windows, materials, etc.), Control Systems, Electrical Systems, etc. Integrated BIM with Authoritative OSi RDF Representations with added geolocation to support geospatial functions
Other Prototype Tools under development
Juma Uplift Visual tool to create and edit uplift mappings from relational to RDF Uses Blocky: mappings created by connecting blocks together Export mappings as R2RML, SML and more. Reference Research Paper and Demo Videos: https://www.scss.tcd.ie/~crottija/juma/
SUMMR Interlink Validation Tool SPARQL Usage for Mapping Maintenance and Reuse SPARQL templates for maintenance and reuse SUMMR Interlink Validation Tool Facilitates template for discovering invalid interlinks From a source dataset to multiple target dataset Used in DBpedia quality activity during annual release process Reference Research Paper: http://www.tara.tcd.ie/handle/2262/77312 Demo Video: https://www.youtube.com/watch?v=iYWaCnQRHpg
With thanks to the Team Ordnance Survey Ireland (OSi) Lorraine McNerney Eamonn Clinton ADAPT @ Trinity College Dublin Rob Brennan Senior Research Fellow Christophe Debruyne Research Fellow Kris McGlinn Research Fellow Alan Meehan Research Fellow Darragh Blake DLAB Research Engineer Aoife Brady DLAB Project Manager