we also found this same phenomenon during the February bush-fires in
Victoria. People were being reported missing - and one way to work out
priority for cases was by number of times a person was reported
missing.

But having said that some way of identifying duplicates - even some
statistical measure of the likelihood of a report being a duplicate
would be useful. A statistical distance between reports that looks at
a number of dimensions like time, spatial & attributes. Damn I wish i
was coming to wherecamp so I work on hacking this

we did write something to measure the closeness-in-time on another
project. you can check it out here

http://github.com/sabman/time-tree/tree/master

This is a digression but an interesting one: Essentially we wrote it
coz we needed it to workout the location of a spot on the sea floor
that was GPS time stamped  and had GPS lat/lon. But this was not the
true location of the underwater camera & there was another file which
contained lat/lon based on underwater acoustics. Problem was that
there was not an exact match b/w GPS time and acoustic beacon.
Time-tree can find the closest match in a time series when give a time
and max time offset. Plus it has a tree structure for storing time so
when searching it can quite quickly converge on the closest time. free
free to fork it :)

_______________________________________________
Geowanking mailing list
[email protected]
http://geowanking.org/mailman/listinfo/geowanking_geowanking.org

Reply via email to