23 Aug 2007 - KML Placemarks as Observations

In GML a distinction is drawn between "authorized" features and observations about features. This distinction is an important one for all kinds of information sharing, and gets at the core of issue so actively debated at GeoWeb 2007 concerning the relative roles of user and "rpofessionally" generated data.

It is clear that users (meaning anyone) can generate geographic information. If say the street light outside my house is out that can indeed be useful information. I might also say that one of my neighbours encroaches on my property and even sketch out where I think the boundaries really are. It would be absurd, however, for me to simply change the boundary, since that is governed by legal processes and can only be changed by such processes.

While I am not authorized to change the official boundary - meaning I cannot change the parcel feature description for the parcel that I own - I can create, declare and submit an observation that "according to me" the boundary is such and such. This I can draw on Google Earth or other such public viewing instrument. It then requires an authorization process to determine where the "true" boundary should be located. Note that this true boundary may be the same as my observation, or may be quite different (I or my neighbours may request a survey and the final result may be an adjustment based on multiple observations).

Think of this in more general terms with respect to KML Placemarks on Google Earth/Maps. When I mark something in GE/GM, what am I doing? I can of course be simply annotating the visual image or map. This can, however, be viewed as creating and declaring an observation - "this is the Lions Gate Bridge and it has three lanes and I noted this at 3:00 am, June 22, 2007". Note that an observation requires the time at which the observing took place, ideally the location of the observer (I was looking from Prospect Point), and the procedure or instrument used in performing the observing. Of course Placemarks were NOT designed with the view in mind, so in most cases you cannot record all of this information, at least not in a machine readable manner (i.e. you can record it in Descriptions).

Placemarks that incorporate photographs make this observation idea even more explicity. Look at a Panoramio photo and see it as an observation which can then be used to create feature information - meaning an authorized view of some entity in the world. The utility of such observations to create candidate feature models has been clear for a long time, but only recently have people been able to mine significant volumes of these pictures to create feature geometry models as demonstrated by Microsoft Photosynth technology.

So how do we go from Placemarks as observations to real authorized features? Stay tuned.

Comments

No comments yet.