3. Geo-reference
Association between a piece of information and a
place
Geo-referenced information is a large part of
whole on line information
Geo-reference can be explicit (geo-tagging) or
implicit
Geo-reference can be at different levels (country,
city, street, lat. and lon.)
5. Extracting “where” from web pages
the extraction of geographic terms from structured
and, more challengingly, unstructured data;
the identification and removal of ambiguities in
such extraction procedures;
Paris Hilton ambiguity
11. Summarization
Difficult task
Image analysis alone is poor at capturing semantic
concepts of an image
In multi user sets the popularity is used but the bias
introduced by few photographers can induce to
overestimate the importance of some objects
12. Summarization
Conceptual framework1:
A real point-of-interest is a point of accumulation for
several photos
A real point-of interest is of interest for many
photographers and not only for few
Textual tags if rarely found elsewhere are likely a
symptom of some peculiarity of a place
1) Alexander Jaffe et al. Generating Summaries for Large Collections of Geo-Referenced
Photographs
20. The test
Panoramio Vs. Flickr Maps Vs. Fractal View
Test the usability
Test by interviews and eye-tracking
Task: find 10 POI to collect in a favorite list
Number of Testers=10
31. 3 different ways
Panoramio: gaze continuously moves from map
controls to thumbnails and vice versa
Flickr: select a cluster and then browse (loosing
geographical details)
Fractal View: select a bounding box and browse,
pictures are tiled with geographical constraints
and move picture-by-picture
32. Conclusion
Geo-tagged items are growing with exponential
pace
Summarizing and visualizing large collections of
geo-tagged objects is an open issue.
Mobile devices equipped with GPS and compass
are a test bench for new interaction design
solutions
33. Acknowledgement
Work part of
Geoweb & Mobile Lab
CRS4, in collaboration with
DistrICT, UniCA, UniSS, Sardegna Ricerche