Geographic information in govt linked data

Geographic Information in Government Linked Data

W3C Government Linked Data F2F 29 June 2011

Todd Pehle

tpehle (at) orbistechnologies.com

Agenda

• The Emerging Web of Linked Locations• The Role of Government SDI

(Spatial Data Infrastructure) in Linked Data• The Role of Lightweight Linked Locations in

Government Linked Data• The Role of Citizen Sensors and VGI

(Volunteered Geographic Information) in Linked Data• Geo Vocabularies, Tools & Best Practices Needed

in Government Linked Data

The Emerging Web of Linked Locations

Orbis Technologies, Inc. 4

Linked Locations

• Linked Data + Location Data = Linked Locations• Web of Linked Locations = a spatiotemporal projection

of the Web of Linked Data (a global spatial KB)

Cartographic Projection

GeoS

PA

RQ

L

Spa

tial

Rea

soni

ngLOD Spatial Projection

Building the World’s Spatial KB

Geo Web has base vector layers (GIS Data) to be connected to Web of Linked Data.

Linked Locations inter-connects Geo Web features & bridges Linked Data with Geo Web.

Linked Data exposes thematic attribution & relationships of Linked Locations.

Linked D

ata + G

eo W

eb

Geo Web Links

(horizontal links)

(vertical links)

The Role of Government SDI (Spatial Data Infrastructure) in Linked Data

Spatial Data Infrastructure (SDI)

• Spatial Data Infrastructure (SDI):

“An SDI is a coordinated series of agreements on technology standards, institutional arrangements, and policies that enable the discovery and use of geospatial information by users and for purposes other than those it was created for.” - Werner Kuhn, IFGI

• Succinctly, SDI is interoperability and sharing of geospatial information• SDI utilizes many OGC service and ISO metadata specifications:

– WMS, WFS, CSW, WCS, GML, ISO19xxx metadata profiles, etc.

• SDI important in government for science, environmental monitoring, emergency response, natural disasters, homeland security, etc.

• SDI spans local, state, federal government plus industry & academia• SDI in U.S. began in 1994; President Clinton signed an Executive

Order calling for the establishment of a NSDI (National SDI)• EU INSPIRE, GEOSS, UNSDI, GMES also examples of SDI

Linked Spatial Data Infrastructure

• Linked Spatial Data Infrastructure (Linked SDI):– SDI data published as Linked Data– Work in progress

• Traditional SDI largely utilized by the GIS domain only• Linked SDI offers the chance to open up rich location data

to a much wider audience• Linked SDI could be consumed just like all Linked Data via

SPARQL (to include GeoSPARQL) and URI dereferencing• Linked SDI could uniformly expose the massive amounts

of structured GIS data into the Linked Data Cloud• Thus the lines between the Semantic Web and Geo Web

could become very blurry!

Examples of SDI (Cited in Wikipedia)

Importance of Government Linked SDI to Linked Open Data

• Governments are HUGE producers/consumers of geo data• Governments as producers have an opportunity to

bootstrap billions of triples into LOD and RDF clouds • Governments as consumers have an opportunity to make

better informed analytic decisions; but also to produce enriched correlations and republish to LOD

• Government Linked SDI could bootstrap LOD with “base layers” of spatial knowledge (NASA, USGS, Census Bureau, Ordnance Survey, INSPIRE, etc.)

• Government Linked SDI can also provide “overlay layers” for certain contexts (EPA, FEMA, HHS, etc.)

• Published data will have varying degrees of complexity; some agencies may simply publish W3C Basic Geo while others may publish heavyweight GIS

Bootstrapping LOD with Government Linked SDI (LSDI)

Government LSDI Base Raster Layers

Government LSDI Core Vector Layers

Government LSDI Contextual Overlays

The Role of Lightweight Linked Locations in Government Linked Data

Lightweight Linked Locations in Government Data

• SDI offers rich data, but traditionally is “heavy” GIS• Often requires specialized knowledge and tools to utilize• However, many agencies have data (unstructured reports

and semi-structured work products) that have simple location information within it:– Placenames– Addresses/Geocoding– Latitude/Longitude fields– GeoTagged Images

• This shouldn’t require GIS departments to publish as LOD• With the rise of Neogeography (DIY mapping) simpler tools

can expose geographic information on the Geo Web• Neogeography tools could also be used to publish

lightweight government Linked Locations data

The Role of Citizen Sensors and VGI(Volunteered Geographic Information)

in Linked Data

Citizen Sensors, NGOs and Linked VGI

• Volunteered Geographic Information (VGI):“The harnessing of tools to create, assemble, and disseminate geographic data provided voluntarily by individuals”

– Mike Goodchild (UCSB)

• User-generated geo content, crowd sourcing, collective intelligence generated by ordinary citizens and NGOs

• Cell phone sensors (cameras, GPS, augmented reality, etc.)• VGI produces “hyper local” and “hyper temporal” knowledge that

complements SDI data• Linked Gov Geo Data can

embrace VGI to produce “hi-res” knowledge of location & events

• USGS held VGI workshop in 2010• OSM volunteers produced

amazing maps for Haitian earthquake first responders

• What if Gov VGI was Linked VGI? VGI in Haiti: Earthquake Damaged Buildings from Open Street Map

Geo Vocabularies & Tools Needed in Government Linked Data

• FOAF = Person• Organization Ontology = Organizations• OWL-Time = Time• Event Ontology = Events (sort of???)• SIOC = Linking online communities, wikis, blogs, etc.• SKOS = Thesauri, classification schemes, taxonomies• Dublin Core = Authoring Metadata• voiD = Vocabulary Of Interlinked Data • W3C Provenance = source metadata

The Lack of Common Geo Vocabularies in the Web of Linked Data

• W3C Basic Geo = 1 Geo Model! (GeoSPARQL soon)

GeoSPARQL

• Draft specification recently released by OGC• Consists of a small vocabulary that defines a geographic

feature and a set of spatial relations• Defines a set of spatial functions based on OGC Simple

Features Specification• Allows powerful query over RDF graphs and geometry• GeoSPARQL vocabulary does not replace W3C Basic

Geo; should be complementary• Represents MAJOR bridge between the Semantic Web

and Geo Web• Represents a key enabler to more uniformly access and

integrate geospatial information in Linked Data

Example GeoSPARQL Query

SELECT ?airportWHERE { ?airport rdf:type :Airport . ?airport :hasPointGeometry [ ogc:asWKT ?aPointGeom ] FILTER(ogcf:distance(?aPointGeom, “POINT(-122.4192 37.7793)”^^ogc:WKTLiteral, ogc:km) <= 100) }

Example Query: Find airports within 100 KM of San Francisco

:SemTech2011 :locatedIn :SanFrancisco .:SanFrancisco :hasPointGeometry [

ogc:asWKT “POINT(-122.4192 37.7793)”^^ogc:WKTLiteral ] .

Example Data: SemTech 2011 is located in San Francisco

Example GeoSPARQL: Features Contained In User-Defined Polygon

GeoSPARQL enables rich spatial queries over RDF

Example Query: Find tourist attractions contained in Capitol Mall area in Washington D.C.

The Linked Locations Core?

GeoSPARQLSpatial Relations

Events

Feature Types

Metadata

Note: Citizen Sensor “Observations” are separate but make use of these core models

Linked Locations Tools & Best Practices

• Almost no tools exist!• Almost no best practices exist!• Need many of the following tools:

– Geo2RDF converters– GeoD2RQ– More GeoSPARQL-enabled triple stores– GeoSPARQL publishing tools– Slippy Map extensions

• Need many of the following best practices:– Geo Provenance publishing– Geo voiD and geo SCOVO publishing– GeoSPARQL & W3C Basic Geo alignment & interplay– Spatiotemporal data publishing best practices (place + time)– Spatial Dataset Change publishing– Linked SDI (OGC-to-LOD) publishing best practices– Linked VGI (citizen sensor) publishing best practices

Thanks!