linked (geo) data - adding a spatial dimension to the web of data

Universität Leipzig ▪ Agile Knowledge Engineering and Semantic Web (AKSW) Authors: Sören Auer, Jens Lehmann, Claus Stadler Slide 1

Linked (Geo) DataAdding a Spatial Dimension to the Web of Data

Claus Stadler, Sören Auer AKSW Research Group University of Leipzig

http://linkedgeodata.org

http://linkedgeodata.org/


Structure

● Linked Data● LinkedGeoData

● Architecture● Mapping to RDF/OWL● Interlinking● Applications● Issues & Future work

● Open Governmental Geo Data


The emerging Web of Linked Data


LinkedGeoData

Conversion, interlinking and publishing of OpenStreetMap.org data sets as RDF.


Motivation

● Ease information integration tasks that require spatial knowledge, such as

● Offerings of bakeries next door● Cinemas nearby and their programs● Historical sights along a bicycle track

● Therefore use RDF/OWL in order overcome structural and semantic heterogeneity.

● LOD cloud contains data sets with spatial features● e.g. GeoNames, DBpedia, US census, EuroStat● But: they are restricted to popular or large entities like countries, famous

places etc.● OSM offers buildings, roads, mailboxes, trash bins/recyling, ...


What do we offer?

● Very large RDF knowledge base derived from OpenStreetMap

● REST API, static & live SPARQL endpoints

● Update propagation of the added and removed triples

● 845.000 Interlinks with DBpedia, GeoNames, and FAO

● LGD-Browser


Architecture


OpenStreetMap - Datamodel

● Basic entities are:● Nodes Latitude, Longitude

● Ways Sequence of nodes

● Relations Associations between any number of nodes, ways and relations.

● Each entity may be described with tags:

key-value pairs, such as (amenity, school)


Example: Leipzig's Zoo


OSM entities → RDF

● Three types of mappings based on (osm-id, key, value)

● Text– (5, name, Leipzig) → lgd:node5 rdfs:label ”Leipzig”

– (5, name:de, Leipzig) → lgd:node5 rdfs:label ”Leipzig”@de

● Datatype properties– (6, seats, 4) → lgd:node6 lgdo:seats ”4”^^xsd:integer

● Classes/object properties– (7, amenity, school) → lgd:node7 a lgdo:School

– (8, amenity, rare value) → lgd:node8 a Amenity

– (9, religion, hindi) → lgd:node9 lgdo:religion lgdo:Hindi


Ontology

● Text, class, and object property mappings manually crafted

● Subclass relations inferred from class-mappings:(..., amenity, school) → ... a School

(..., amenity, rare value) → ... a Amenity

School subClassOf Amenity

● Automatically mapped OSM keys with mostly numerical values to datatype properties● boolean, integer, float


Ontology

● Enriched classes and properties with multilingual labels from TranslateWiki● http://translatewiki.net

● Imported icons for 90 classes from the freely available icon collection from the SJJB Management● http://www.sjjb.co.uk/mapicons/

http://translatewiki.net/

http://www.sjjb.co.uk/mapicons/


Ontology


Instance Data


Statistics (2011-Feb-23)

● 222.539.712 Triples● 6.666.865 Ways● 5.882.306 Nodes

● Among them● 352.673 PlaceOfWorship● 60.573 RailwayStation● 59.468 Recycling● 50.955 Town● 30.099 Toilet● 7.222 City


DBpedia Mapping

Given a point from DBpedia, GeoNames, …, query LGD points with certain types within a maximum distance

Basic idea (performed with Silk):

● Find most similar type(s) in LGD for a type in another knowledge base

● Compute spatial score

● Compute name similarity (rdfs:label)

● Combine both scores

● Depending on final score, either automatically accept/reject links or mark for manual verification.


How to access the data?

● Rest Interface (based on Postgis DB, full osm dataset loaded, > 1billion triples)

● Supports limited queries (e.g. circular/rectangular area, filtering by labels)

● Sparql Endpoints (based on Virtuoso DB, subset of osm dataset loaded, ~222mio triples)

● Static (http://linkedgeodata.org/sparql)

● Live (http://live.linkedgeodata.org/sparql)

● Downloads (same data as in the SPARQL endpoints, http://downloads.linkedgeodata.org)

http://live.linkedgeodata.org/sparql


Applications: LinkedGeoData Browser

http://browser.linkedgeodata.org

http://browser.linkedgeodata.org/


Applications: Generic spatial data browsing widgets


Applications: MovieGoer

● MovieGoer– Scrapes websites with information

about the programs of cinemas in Innsbruck and Munich

– Interlinked with LinkedGeoData in order to obtain address information

– During development: missing cinemas added to OSM; became available for interlinking through LGD-Live.

http://lokino.sti2.at/

http://lokino.sti2.at/


More Applications

● Layar

Augmented reality browser for mobile phones, features a LGD layer

http://layar.com

● Vicibit

Generates embeddable HTML code for displaying a map of POIs of certain types in a specified area

http://vicibit.linkedgeodata.org

http://layar.com/

http://vicibit.linkedgeodata.org/


Issues & Future work

● Converting all OSM data to RDF not feasible:

> 1.000.000.000 nodes

→ only a subset of OSM's data available via the SPARQL endpoints

● Virtuoso only supports point geometries

● Possible solution: Rewriting SPARQL queries to the OSM SQL schema● Work in progress, no estimate on a release yet


Issues and Future Work

● Investigating further applications of LGD, such as using the LGD knowlegde base for named entity recognition/resolution


Open Governmental Geo Data

● Most governmental data has a spatial dimension● transportation, statistics, finances, etc.

● Geo Data is a driver in the Open Government Data realm● ordenance survey, geodata.gov.gr, …

● It is still difficult to extract, publish, interlink, author and visualize Governmental Linked Geo Data● LOD2 Stack (ckan integration, browsing widgets,

integrated comprehensive tool compilation)


Best practices

● Reuse of vocabularies

● vCard: address information, phone & fax

● Good Relations: Opening hours, product descriptions

● GeoOWL: latitude, longitude, polygons

– GeoSPARQL Ontology● Foaf: based_near

● Dublin core (dc): Document meta data

● …

http://www.leipzig.de/de/buerger/jugend/betreuung/kitas/

http://www.leipzig.de/de/buerger/jugend/betreuung/kitas/


Transportation


Comprehensive Knowledge Archive Network (CKAN)


Financial Transparency


The End

Thanks for your Attention!

linked (geo) data - adding a spatial dimension to the web of data

Technology

semantic web aksw authors

jens lehmann

claus stadler slide

leipzig lgd

semantic web akswauthors

web of data http

large rdf knowledge

school lgd