Using GPS Tracks as Contextual Metadata for Multimedia
In the reading “Context Data in Geo-Referenced Digital Photo Collections”, a single GPS point is used as metadata for an image. In fact, GPS device could capture more than just a point; it could capture the whole track of a user’s journey, and using the GPS tracks as metadata may be more interesting than using point alone. The image below shows a trip around Tian’anmen in Beijing, with images connected together by a GPS track.
There are at least two benefits about using a GPS track, instead of a point:
1.Better visualization. Users could “replay” the whole journey on a map together with images in animation (which I implemented myself with Javascript and Live Map API)
2. GPS tracks reveal more about users’ behavioral patterns, and these patterns can be used to improve mobile/location search. On an individual level, car drivers care less about distance than pedestrians, and traffic mode can be inferred from GPS tracks using supervised learning [1]; on a social level, locations frequently visited could be regarded as more “popular” and thus be given higher ranking in local search. Rich geographic information in GPS tracks is invaluable to mobile and location-based IR.
[1] “Learning Transportation Mode from Raw GPS Data for Geographic Applications on the Web”, WWW2008