Discussion Paper

Populating <dc.coverage> - Discussion Paper

Scope

Dublin Core, selected as the metadata standard for documenting Government of Canada Web resources, is a discovery-level metadata standard, capable of leading users to a resource. Dublin Core metadata cannot be expected to support resource evaluation or resource access. We must find a way to utilize the Dublin Core "Coverage" element to maximize its potential for discovery, when searching by location on Government of Canada Web sites. This discussion paper presents several means of achieving this.

Introduction

The Geographic Coverage Sub-group of the GOL Metadata Working Group met on December 18, 2001. The objective: finding methods of populating the Dublin Core "Coverage" element to enable geospatial searching on GOL Web pages by GoC search engines.

There was notable interest amongst the group on the following:

• adhering to international standards (Dublin Core Metadata Initiative (DCMI), ISO 19100 Geographic Information and Geomatics);
• a transparent method of providing coordinate geometry based on user input from controlled vocabularies where desired or required;
• providing commonly known geographic identifiers, including "regional" geographic identifiers with "fuzzy" boundaries, such as "the Prairies", "the Rockies", and so forth.

This draft document, provided by GeoConnections and GeoAccess Division, Natural Resources Canada, is intended to stimulate further discussion within the Geographic Coverage Sub-group and the GOL Metadata Working Group. Its purpose is to suggest methods of populating "Coverage" using known vocabularies that will enable efficient searches, and leave the door open for potential Web services that may be used to further utilize this information. A number of possible named geographies, source vocabularies and HTML encoding are included. References for further investigation into Web services and related topics are also provided.

GeoConnections, a federal program led by Natural Resources Canada, is developing the Canadian Geospatial Data Infrastructure (CGDI). The CGDI is intended to provide Canadians with access to geospatial information, technologies and services through a network of data, service and technology suppliers found within federal, provincial and territorial governments, the private sector and the academic community. Please visit http://www.geoconnections.org/CGDI.cfm/fuseaction/aboutGcs.welcome/gcs.cfm
home.welcome/gcs.cfm/about/about_ataglance.html for more information on GeoConnections.

Populating Coverage

A proposal for populating <dc.coverage> with a geographic point or bounding box scheme was put forward in 2001. During the December 18 meeting, the Geographic Coverage Sub-group discussed using geographic identifiers, like place names or postal codes, as another input for <dc.coverage>. Using geographic identifiers for this element increases the accessibility/usability of <dc.coverage> for a wider (non-geospatial) audience, who may not be comfortable with lat/long coordinates. This discussion led to an action item: identify and define a number of named geographies that could be included in the metadata, in addition to, or instead of, point and bounding box. Named geographies from known, controlled sources can potentially be translated to geographic coordinates via Web services, such as a gazetteer service.

The Coverage element is repeatable. Consequently, the metadata may contain any combination of coordinate geometry and/or geographic identifiers. The use of one or more geographic identifier may bring improved search results. A point location cannot define an area, and a bounding box often captures more information than required within its extents. Geographic identifiers may be used to sharply define a location if translated to coordinates, by deriving a polygon, or tiling a collection of smaller bounding boxes to form clusters. A place name or geocode is not a simple character string, it is a key which can unlock rich information that may be used to search more effectively. This approach is also being used in the Canadian profile of the ISO 19115 Geographic Metadata standard, in which both geographic coordinates and geographic identifiers are considered valid means of defining a location.

Named geographies

"Content data for some elements may be selected from a "controlled vocabulary," which is a limited set of consistently used and carefully defined terms. This can dramatically improve search results because computers are good at matching words character by character but weak at understanding the way people refer to one concept using different words, i.e. synonyms. Without basic terminology control, inconsistent or incorrect metadata can profoundly degrade the quality of search results." - DCMI Usage Guide

A great number of named geographies could be used to populate <dc.coverage>, and each department/cluster has its own requirements.  It is important to narrow the choices to promote ease of use and comprehension. This is achieved by "typing" named geographies into several well-chosen and well-defined categories. The use of types standardizes input to services and allows for semantic interoperability. The OGC refers to this as a "Geocoding Type" and defines it as follows: "A Geocoding Type is a Universal Type that can be understood by any application, developer, or reader of the specification. It is the common international agreement and the first common basis for interoperability." ISO 19112 - Geographic Identifiers calls them "Location types" and defines them as "A spatial reference system using geographic identifiers comprises a related set of one or more location types, together with their corresponding geographic identifiers."

The various "types" suggested for use by the Geographic Coverage Sub-group include postal codes, street addresses, geographic regions and administrative areas. Geographic identifiers must be drawn from known (registered) sources that may offer other geospatial attributes, preferably coordinate geometry. This means controlled vocabulary or code lists must be found or created, and metadata populated according to the offered vocabulary. In accordance with GOL requirements,

"The use of controlled vocabulary is necessary to classify and describe information, and to support navigation, searching, information sharing and interoperability goals of Government On-Line (GOL). Controlled vocabulary used by the Government of Canada should be registered and publicly available." - GOL Controlled Vocabulary Standard (TBITS 39.2).

Restricting the Coverage element to a certain number of named geographies (such as those described below) will introduce order. Without controlled vocabularies, the element may be populated in a haphazard fashion that will lessen the value of the metadata. Whether services are utilized to translate geographic identifiers to coordinates or not, Coverage needs to be populated in a consistent manner, as this will result in successful searches by Government of Canada search engines.

Semantics of Named Geographies

Below find semantic definitions for a number of "geocoding" or "location" types, along with suggested vocabularies and examples.
(partially adapted from OpenGIS Consortium Discussion Paper 01-026r1 "Geocoder Service Draft Candidate Implementation Specification 0.7.6", March 28, 2001)

Type: Geographic Region
Semantic Definition  (for group comment) - a commonly known area that does not necessarily conform to administrative area boundaries. A region, a community, a place.

Suggested Source of Controlled Vocabulary:
Title: "Regions of Canada"
Originator: The Atlas of Canada
Description: In order to satisfy a requirement from HRDC for a short list (under 50) of regions across Canada, a list was produced that defines regions based upon census divisions. The list consists of six large interprovincial regions and 39 intraprovincial regions. Since the regions are based upon CD's or provincial/territiorial boundaries, they are also defined by geographic coordinates.
Cost: Freely available.
Note: Not yet registered. See Appendix C for example.

Suggested HTML encoding:
<meta name="dc.coverage" scheme="gcregions" content="Prairie Provinces">

Type: Administrative Area
Semantic Definition (for group comment) - an area with legally defined boundaries, such as a province, county, electoral zone, census division.

Suggested Source(s) of Controlled Vocabulary:
Title: The Canadian Geographical Names Database (CGNDB) 1997
Originator: Geographical Names Board of Canada
Description: Land Districts, Geographic Counties, Section-Township-Range, Indian Reserves, Military Areas, Conservation Areas, Water Features, Provinces and Territories. The new version of the GeoNames database will also have an API (http request, XML response) and will eventually be linked into feature extents (e.g. bounding box, bounding polygon), in addition to the existing centroid. This project is just getting underway, and will be a valuable future resource.
Cost:

Title: ISO 3166 - Codes for the Representation of Names of Countries and Their Subdivisions
Originator: ISO
Description: Codes representing international place names. 3166-1 provides country codes, 3166-2 provides country subdivision codes, and 3166-3 provides codes for formerly used names of countries.
Cost:

Title: Federal Electoral Districts of Canada
Originator: Elections Canada, Electoral Geography Division
Description: 301 Canadian Electoral Districts. Description of redistributed district boundaries will be available online in March 2002.
Cost: Freely available on http://www.elections.ca
Note: See Appendix C for example.

Title: GeoBase Administrative Boundaries, Canada
Originator: Natural Resources Canada, Earth Sciences Sector, The Atlas of Canada
Description: Provinces and territories of Canada. Reliable framework data for the practical use of geospatial data.Restructured, improved and generalized from VMAP Level 0 data. Polygon defined for all provinces and territories in Canada.
Cost: Available free of charge from GeoGratis.
Note: See Appendix C for example.

Title: Administrative Boundaries, Canada
Originator: Natural Resources Canada, Earth Sciences Section, Mapping Services Branch, Centre for Topographic Information (Sherbrooke)
Description: The initial version of the Administrative Boundaries file contains six types of entities: international boundaries, provinces, municipalities, national parks, national wildlife areas and migratory bird sanctuaries. A total of 13,351 files at the 1:50 000 scale and 980 files at the 1:250 000 scale are created for the entire Canadian landmass.

Suggested HTML encoding:
<meta name="dc.coverage" scheme="CGNDB" content="Ontario">

Type: Municipality
Semantic Definition (OGC Geocoder) - The name of a city - an inhabited place, a town, village, etc.

Suggested Source of Controlled Vocabulary:
Title: The Canadian Geographical Names Database (CGNDB) 1997
Originator: Geographical Names Board of Canada
Description: City, Town, Village, Hamlet, Unincorporated Area, etc. Place names with a number of associated attributes, including Province/Territory, feature type, location (county or other administrative area) latitude and longitude (point), NTS map and CGNDB unique identifier.

Suggested HTML encoding:
<meta name="dc.coverage" scheme="CGNDB" content="Ottawa">

Type: Municipal Subdivision
Semantic Definition (OGC Geocoder) - The name of a division of a municipality - district, precinct, section, sector, area, neighbourhood, vicinity, etc.

Suggested Source of Controlled Vocabulary: not yet defined.

Suggested HTML encoding:
<meta name="dc.coverage" scheme="unknown" content="Ottawa South">

Type: Street
Semantic Definition (from OGC Geocoder) - The street address or name, such as 5^th Avenue, Avenue de l'Observatoire, etc.

Suggested Source(s) of Controlled Vocabulary:
Title: Block-Face Data File (BFDF) (1996 Census), Canada
Originator: Statistics Canada, Geography Division
Description: The Block-Face Data File (BFDF) contains 1996 Census population and dwelling counts for block-faces in urban centres covered by the Street Network Files (92F0024XDE). A block-face is generally one side of a city street between two consecutive intersections; it is also the smallest geographical unit available from Statistics Canada. The BFDF also links the block-face to all other levels of standard geography (enumeration areas and above) through geographic codes. The file includes street names with address ranges as well as x, y (latitude/longitude) coordinates for a point representing the approximate centre of each block-face. The block-face information includes street name (including street type and direction), address range, geographic codes, x,y coordinates for a representative point and 1996 population and dwelling counts.
Cost: Custom orders available upon request.

Title: Framework Road Data Model
Originator: Transportation Node/Framework Activity/GeoBase, Elections, OMNR and CTI
Description: Based on GDF (?) and includes road names and address ranges as well as coordinate geometry.
Cost: Free - Geobase data

Suggested HTML encoding:
<meta name="dc.coverage" scheme="unknown" content="Coldstream Drive">

Type: Street Number
Semantic Definition (from OGC Geocoder) - The number of the location in the street, or at the place name. It can be a house number (128, 4B)

Suggested Source of Controlled Vocabulary:
Title: Block-Face Data File (BFDF) (1996 Census), Canada
Originator: Statistics Canada, Geography Division
Description: The Block-Face Data File (BFDF) contains 1996 Census population and dwelling counts for block-faces in urban centres covered by the Street Network Files (92F0024XDE). The file includes street names with address ranges as well as x, y (latitude/longitude) coordinates for a point representing the approximate centre of each block-face. The block-face information includes street name (including street type and direction), address range, geographic codes, x,y coordinates for a representative point and 1996 population and dwelling counts.
Cost: Custom orders available upon request.

Suggested HTML encoding:
<meta name="dc.coverage" scheme="unknown" content="55">

Type: Postal Code
Semantic definition (from OGC Geocoder)- An area encompassing a sections of a street, a collection of streets, an establishment, a structure, a group of post office boxes, and so forth, which is assigned the same code by the Postal Service of the country.

Suggested Source of Controlled Vocabulary:
Title: Postal Code Conversion File (PCCF), Canada
Publisher: Statistics Canada, Geography Division
Description: The Postal Code Conversion File (PCCF) provides a link between six-character postal code and standard 1996 Census geographic areas (such as enumeration areas, municipalities, census tracts, and so on). It also provides the x,y (latitude/longitude) coordinates for a point representing the approximate location of the postal code to support mapping. The Postal Code Conversion File (PCCF) is a digital file, which provides a correspondence between the six character postal code and Statistics Canada's standard geographical areas for which census data and other statistics are produced. Through the link between postal codes and standard geographic areas, the PCCF permits the integration of data from various sources.
Cost: Canada coverage $9000.00, subsequent updates $1500.00

Suggested HTML encoding:
<meta name="dc.coverage" scheme="unknown" content="K0A3P0">

(Previously suggested):

DCMI Point
Semantic Definition - identity of a place using a point location, described using coordinates in an identified Cartesian system.

HTML Encoding:
<meta name="dc.coverage" scheme="DCMIPoint" content="north=49.0;east=-80.0>

DCMI Box
Semantic Definition - identity of a place defined by a minimal rectangular box which fully encloses the place, whose faces are aligned parallel with the axes of an identified Cartesian coordinate system.

HTML Encoding:
<meta name="dc.coverage" scheme="DCMIBox" content="northlimit=77.0;southlimit=-70.0;westlimit=69.0;eastlimit=61.0">

Further Information:

Appendix A - Supporting Services, which could be provided by GeoConnections

Geocoder Service

"A Geocoder Service is a network-accessible service that transforms a description of a feature location, such as a place name, street address or postal code, into a normalized description of the location, which includes a coordinate geometry...this service will determine the geometries for one or more features, given their associated well-known feature location descriptions, which are specified to the service at run-time, through a query." OGC Discussion Paper 01-026r1

Gazetteer Service

"A Gazetteer Service may be thought of as a special case of a Geocoder Service that is an authority for place names and their associated geometries." OGC Discussion Paper 01-036

The Canadian Geospatial Data Infrastructure (CGDI) team are currently developing the CGDI Gazetteer service. Links to a presentation on this work in progress are given below.

Scenario document:
http://www.geoconnections.org/
devNetwork.meetings/pgm_id/25/gcs.cfm

Appendix B - Sample Gazetteer Query/Response

Accessing geographic properties of a known feature (from OGC Gazetteer)

<?xml version="1.0" encoding="UTF-8"?>
<GetFeature xmlns:gml="http://www.opengis.net/gml"
xmlns:fo="http://www.w3.org/1999/XSL/Format" maxFeatures="3">
<Query typeName="GNS_GAZ">
<Filter>
<FeatureId fid="GNS_GAZ.-474536" />
</Filter>
</Query>
</GetFeature>

<?xml version="1.0" encoding="utf-8" ?>
<gml:featureCollection scope="http://benoit/servlet/wfs/GNS_GAZ?"
xmlns:xlink="http://www.w3.org/1999/xlink"
xmlns:gazetteer="http://www.opengis.net/namespaces/gazetteer"
xmlns:gml="http://www.opengis.net/gml"
xmlns:xsi="http://www.w3.org/2000/10/XMLSchema-instance"
xmlns:wfs="http://www.ionicsoft.com/wfs">
<gml:boundedBy>
<gml:Box srsName="EPSG:4326">
<gml:coordinates>34.416668,29.691668
34.416668,29.691668</gml:coordinates>
</gml:Box>
</gml:boundedBy>
<gml:featureMember>
<wfs:GNS_GAZ fid="GNS_GAZ.-474536">
<gml:description>Garf el Themed</gml:description>
<gml:name>Garf el Themed</gml:name>
<gazetteer:featureDesignation>RDGE
</gazetteer:featureDesignation>
<gml:location>
<gml:Point srsName="EPSG:4326">
<gml:coordinates>34.4166667,29.6916667,0.0</gml:coordinates>
</gml:Point>
</gml:location>
<gazetteer:sourceFeature
xlink:type="locator"></gazetteer:sourceFeature>
<gazetteer:BT xlink:type="locator"
xlink:href="/servlet/wfs/GNS_GAZ?request=GETFEATURE&amp;TYPENAME=GNS_
GAZ&amp;FEATUREID=GNS_GAZ.-451924"></gazetteer:BT>
</wfs:GNS_GAZ>
</gml:featureMember>
</gml:featureCollection>

Appendix C - Sample Controlled Vocabularies

Title: Canadian Provinces and Territories - Bounding Box Coordinates
Maintained by: The Atlas of Canada

Title: Canadian Provinces and Territories Names, Symbols and Abbreviations in English and French
Maintained by: http://geonames.nrcan.gc.ca/english/prov_abr.html

Title: Regions of Canada
Maintained by: HRDC(trusted authority)/Natural Resources Canada (consulting).

Title: 1996 Representative Order - Electoral District
Maintained by: Elections Canada

Note: The following are electoral district of Newfoundland and Labrador, only seven of 301 Canadian electoral districts.