I'd be particularly wary of inferring anything from the EXIF data,  especially the time.

I have a cheap digital camera which is pretty good except that the clock periodically resets to a default time.  I've got a somewhat more expensive digital camera which has the same problem.  I have an android tablet that I assume gets the time from the net and/or GPS,  but when I took it ought of my gym bag the other day I noticed the time display had been switched to 24hrs and the time zone was switched to central.

When I am in the photography habit,  I keep the clock set on my cameras.  Sometimes I fall out of the habit but something interesting happens and you'd better believe I am not going to waste time setting the clock if I get a chance to photograph a burning car!

Similarly when travelling I might be bothered to set the timezone or not,  more likely not if I have a layover in some place like Frankfurt or Schiphol airport.

If somebody decided just to set the clock to Zulu I wouldn't blame them.

Also,  efforts to infer stuff from the EXIF data such as "did the flash go off?" rarely produce interesting results.  For instance,  it's a good habit to use the flash when you take photos of people outdoors on a bright day because it softens the shadows.  Some people do it all the time and the auto mode on some cameras does it by default too.  Thus,  the flash is not an indicator that a photo was taken at night,  indoors,  in the dark,  etc.

If you filter on things like that,  or the ISO level,  or the exposure,  or aperture,  you're unlikely to get categories that are useful.

On Wed, Aug 20, 2014 at 7:52 AM, Markus Krötzsch <markus@semantic-mediawiki.org> wrote:
On 20.08.2014 10:46, Gerard Meijssen wrote:
When I add statements with "is a list of", the item I refer to works as
a base. It and all subsequent statements are required to be the result
of the result that is generated by WDQ in the background. The results
are shown automatically from within Reasonator.

The hack is in having Reasonator interpret the limited expressions
available. Then again, calling Reasonator a hack is a disservice to the
real application it provides.

Not sure what you refer to, but there might be a misunderstanding here. I was using the word "hack" in my email to refer to the proposal of using additional qualifiers to express queries in Wikidata. That was a new proposal in the email I replied to and had nothing to do with Reasonator or your annotations.


Wikidata-l mailing list

Paul Houle
Expert on Freebase, DBpedia, Hadoop and RDF
(607) 539 6254    paul.houle on Skype   ontology2@gmail.com