location-formats
Location Research
This is a list of previous efforts at location / geographic ("Geo") data formats and protocols, as background research for developing a location format.
Authors
- Tantek Çelik
- Bud Gibson
- Ryan King
- ...
- please add yourself if you help(ed) with this document.
Location Examples
What People are Publishing
- links to mapping services, canonical URLs to 3rd party services like:
- Google Maps
- Map Quest
- Yahoo Maps
- human / named / legislated formats
- named places, e.g. Westin St. Francis
- addresses, including zip codes
- global/mathematical/geometrical
- Lat & Lon
- MAKE has a GeoURL button
- Geo in RSS. Means a particular blog post is associated with a specific lat/long. Not specific about whether it means the location that the post was made from or the location was the post was about.
- Lat & Lon
Location Formats
vCard & hCard
vCard (RFC2426) and hcard include 'adr', a way to markup addresses. They also have a field called 'geo' for lat/long.
UPU S42 & OASIS xNAL
UPU S42 is a standard for representing postal addresses. UPU, the Universal Postal Union, is the consortium of all national posts. See UPU S42 Announcement. xNAL, the XML Name and Address Language, is an earlier attempt to standardize on the components of addresses developed by OASIS, a consortium of business-process weenies. The distinction between vCard/hCard and S42/xAL/xNAL comes down to whether the address line elements themselves are decomposed. For example, in xNAL you can specify components like street number, street prefix directional, street name, street type, street postfix directional, subaddress type (Suite), subaddress number.
It is not always possible to conclusively determine these components from an address line, even in the US where addresses are more canonical than others. "A 4TH ST W STE 10" is a legal address line in the US. The street number and name are required (in the US) so this could be canonically decomposed as streetnumber="A", streetname="4TH", streettype="ST", postdirectional="W", subtype="STE", subtypenumber="10".
UPU S42 and OASIS xNAL standards are likely overkill for tagging general web content. However, they are important standards that should be considered in web-services involving both geocoding and reverse-geocoding. Perhaps standards related to decomposed addresses is something that should be placed on another page.
In addition, there should be pointers to national address standards.
USPS publishes a complete list of canonical abbreviations for both streettypes and subtypes. Their publications are self-inconsistent, but it is easy to infer the correct mappings. See USPS Acronyms & Abbreviations
New Zealand apparently adopted a variant of xNAL NZ xNAL Guidelines Release 1.0.
ICBM
One comma seperated pair of lat/long <meta name="ICBM" content="XXX.XXXXX, XXX.XXXXX">
[1]
GeoUrl
- GeoURL
- This seems to have a decent amount of geeky adoption, though there are lots of typical invisible metadata problems, coordinates reversed, postive instead of negative etc.
- Multimap.com provides similar functionality to GeoUrl, searching for websites and weblogs which are geotagged or ICBM tagged, and adding them to local information databases.
Flickr Geotags
- Flickr GeoTagging A Greasemonkey script for adding lat and long tags to flick pictures. He uses Google Maps to get the lat and long. Users have to enter search terms (address, etc) and then select a point on the google map. Photos are tagged with "geo:lat=xx.xxxx", "geo:lon=xx.xxxx" and "geotagged" and aggregated at [2]
- A Flickr Group related to the topic
- A visual walkthrough of the technique.
- As of today (2005-06-01), 14k photos have been geotagged on Flickr.
- mappr, mapping flickr also support these geotags.
- flickr also supports EXIF headers, which can be used for storing location.
- plazes also allows flickr photos to be tagged with geotags
This is somewhat of a problem though, as mechanically generated "tags" are not really tags. Including/setting Flickr "Geotags" could be considered pollution of truly user entered text, since the text of geotags is merely an encoding for a point on a map.
Geotagging
But what do these references to "geotagging" mean?
RSS
Several people have tried putting geographic information into RSS.
- This site uses icbm coordinates and has plugins for several blogging packages.
- BlogMapper - appears to be adding a namespace to RSS.
- BlogMap - picks up RSS extension and meta tag values
- worldKit RSS docs - worldKit advocates and understands various flavors of geocoded RSS; particularly the "geo" namespace.
- USGS earthquake feed - most widely used geocoded RSS feed
- Polygons and lines in polygons - A simple format for expressing polygons and lines is defined for worldKit
- Yahoo Maps API - uses "geo" ( http://www.w3.org/2003/01/geo/wgs84_pos# ) and "ymaps" ( http://api.maps.yahoo.com/Maps/V1/AnnotatedMaps.xsd ) namespaces to spatially-reference RSS as part of its API, specifically geo:lat, geo:long, ymaps:Address, ymaps:CityState, ymaps:Zip
RDF
- Geosemantics Interest Group
- GeoOnion - a SW vocuabulary for relating items by distance from each other.
- RDFMap
- GeoRDF similar effort to collect prior art, for an effort in interoperability between OGC standards and various lightweight geo formats.
GML
Geographic Markup Language provides the most complete XML descriptiono of geographic information available. Fortunately, it is also in wide-spread use, so a commonly-used subset of the 600-page specification has emerged. GML is used in the OGC WFS payload described below.
Open Geospatical Consortium - OGC
Since 1994, Open Geospatial Consortium has been developing open specifications to enable the exchange of geographic information between applications. While some GIS vendors opposed OGC initially, essentially all commercial GIS software companies have broken down their proprietary "stovepipes" by embracing OGC specifications. The recent groundswell of ajax mapping applications can benefit tremendously from using OGC specs.
WMS
The OGC Web Map Service (WMS) specification makes it easy to request map images from a map rendering engine, such as the UMN Map Server. It is straightforward to build dynamic tiling ("slippy map") ajax applications that pull in OGC map tiles. At Where 2.0, MetaCarta demonstrated such an OGC-capable ajax GIS client. The portal map providers (yahoo/gmaps/msn/map quest) do not yet offer WMS interfaces, so someone should wrap their proprietary interfaces in a WMS wrapper to encourage them :-)
WCS
Web Coverage Service extends WMS to enable attribute information about large area overview maps.
WFS
The OGC Web Feature Service (WFS) specification is the best way to express geographic information about Web content. It is more complex than the ad hoc geoURL syntax, and for good reason: it can be loaded directly into any OGC-capable GIS even if the coordinate information is in a different projection/datum or is more complex than just a point.
WFS provides a robust means of querying for geographic entities, such as points, polygons, and more complex feature types. One can define new feature types simply by describing them. One can even specify the visual appearance of the geographic features using the OGC Styled Layer Description (SLD) specification.
WFS is to vector map data, as WMS is to raster map data.
Geographic search results, such as the yellow page information from the portals, could be presented in WFS. John Battelle commented that RSS forces publishers to attach their business model to their content, instead of to the web site holding the content, so that they can make money no matter where their content flows. The search portals will probably be forced to do the same, e.g. putting click-through ads in the RSS search results or WFS yellow page results, so that even when you display them in your customized ajax mapping application, the portal publishing the search results can still make money.
Google Earth XML
20050629 at Where 2.0 conference:
"Google Earth has an XML schema for describing a place on the earth. Will be releasing today or tomorrow." - John Hanke of Google/Keyhole.
Official KML Docs and Tutorial
[3] (deprecated now that KML 2.0 is officially released)
NMEA
Standard and Proprietary NMEA Sentances
Wikipedia
Geo:Coordinates in Wikipedia Project
Other Formats
- GPX = GPS XML format
- Wikipedia markup styles
- OpenGuides and evnt are both looking into real-world/folksonomic location name mappings
Location APIs
Google Maps API
Yahoo Maps API
Location Data, Tools and Sites
Address2Coordinates
These systems are freely available sources of address to lat/long mappings (US only). The first few are based on US Census data, so they might be up to date.
- TIGER
- FIPS
- geocoder
- You can get coordinates from UK post-codes from streetmap
- Worldwide city level location data (over 4 million entries) is available from the GNS. This is accessible as a REST service from the worldkit geocoder
- mysociety.org is providing an interesting name-based location look-up services. The results are returned as a CSV file! For instance, the query http://gaze.mysociety.org/gaze-rest?f=find_places&country=US&state=CA&query=sunnyvale returns
"Name","In","Near","Latitude","Longitude","State","Score" "Sunnyvale","Santa Clara County","","37.36889","-122.03528","CA","100"
but with a mime-type (text/csv) that browsers don't understand.
A2B
A2B is a location based search engine. Lat and long only. Seems to pick stuff up from geo meta tags. Pages can either be manually entered or pinged. They support lat/long and ICBM meta tags.
Plazes
Plazes is a system that does geo-info based on the network access point that the user's computer is connected to. They use tagging to identify locations.
Mapping Sites
Location Brainstorming
A lot of was discussed and figured out at the Geo Microformat BOF at O'Reilly's Where 2.0 conference.
See the geo-bof-2005-06-30 minutes/notes.
One of the thoughts that came out of BOF was the idea of location-tagging.
Conclusions on new microformats
The consensus conclusions from the Geo Microformat BOF at Where 2.0 was to define two new microformats to reflect the two kinds of most commonly published geo data.
- adr - for representing / marking-up / publishing address location information. "adr" is simply the hCard "adr" property and all its sub-properties.
- geo - for representing / marking-up / publishing latitude longitude location information. "geo" is simply the hCard "geo" property and all its sub-properties, with the explicitly specified datum of WGS84.