data model for location information

The aim here is to extend the delivery context ontology with 
information on the device's location. This is something we can do 
without having to make any commitments at this stage on specific 
APIs or on privacy issues associated with location information.

Some relevant existing work is:

  * JSR179 - location API for J2ME
  * GPX - common XML format for exchanging waypoints
  * EXIF metadata format for JPEG images

The most essential data is the latitude and longitude
which can be expressed as floating point numbers.

  - Latitude is a number in the range [-90.0, 90.0]

  - Longitude is a number in the range [-180.0, 180.0]

GPS-based sensors may also be able to determine the altitude.
This is represented as a floating point number denoting the
height in metres above the WGS84 ellipsoid.

Estimates of the accuracy of the measurement may be provided.
A simple approach is to provide estimates of the horizontal
and vertical accuracy in metres, as per JSR179.

The location measurement may be associated with a timestamp
and this can be represented in the ISO datetime format,
e.g. 2007-06-10T01:38:55Z

Further possibilities include the observed speed and course
based upon a sequence of measurements, and the orientation,
e.g. the compass azimuth, pitch and roll:

  - Speed: metres per second

  - Course: degrees relative to true north

  - Compass azimuth: horizontal compass azimuth in degrees
    relative to either magnetic or true north

  - Pitch: device tilt in degrees defined as an angle in the
    vertical plane orthogonal to the group and through the
    logitudinal axis of the device

  - Roll: device rotation in degrees around its own
    longitudinal axis

  - A boolean flag indicating whether the compass reading
    is with respect to magnetic north or true north.

New kinds of cameras could include GPS sensors for the location,
magnetic sensors for the compass azimuth, and tilt sensors
for the pitch and roll. This information could be recorded
as part of the EXIF metadata included with each image.

Such information could also be used to select points of
interest to show to the user, and for games as exemplified
by the Nintendo wii.

It is harder to deal with semantically richer information on
device location, e.g. street address, as the schema for such
data varies from one country to another, and moreover it
complicates how you deal with measurement inaccuracies.

I therefore suggest that we don't include such information in the 
ontology at this stage.

  Dave Raggett <dsr@w3.org> http://www.w3.org/People/Raggett

Received on Thursday, 29 November 2007 13:33:49 UTC