5

Now that we are in public beta, we can be queried against using the SE Data Explorer. I've noticed that some software is the answer to man different questions. For example IrfanView.

With a little bit of simple natural language processing, (probably some hacky regexs) it should be possible to determine which product is being recommended in each answer, and thus to run a query against all answers to find which products show up multiple times.

I think it would be pretty cool.

Further things we could do, once we have a programitic way of identify programs the answer speaks of is generate (on say users personal websites, not on SE) Program Lists, linking to there recommendations. Eg list of Image Editors.

2
  • I've thought of making something that would look for links to the same domain repeated multiple times, that would probably work here.
    – Undo
    Feb 24, 2014 at 14:28
  • That proably would yes, because most links that are not to programs aren't going to show up more than once.
    – Frames Catherine White
    Feb 24, 2014 at 14:29

1 Answer 1

Reset to default
7

I was kinda bored, and thus decided to try to do this. I couldn't find an easy way to do it with SEDE, so I hacked together a thing in Python. It's on GitHub.

Here are the results (this can, and probably will, be made better):

6 x http://notepad-plus-plus.org/
6 x http://www.gimp.org/
5 x http://handbrake.fr/
5 x http://www.autohotkey.com/
5 x http://www.ffmpeg.org/
5 x http://www.irfanview.com/
5 x http://www.jetbrains.com/idea/
5 x http://www.pdflabs.com/tools/pdftk-the-pdf-toolkit/
5 x http://www.sublimetext.com/
5 x http://www.virtualdub.org/
4 x http://avidemux.sourceforge.net/
4 x http://brackets.io/
4 x http://en.wikipedia.org/wiki/Emacs
4 x http://getgreenshot.org/
4 x http://gitlab.org/
4 x http://jdownloader.org/
4 x http://www.chiark.greenend.org.uk/~sgtatham/putty/
4 x http://www.code42.com/crashplan/
4 x https://filezilla-project.org/
4 x https://www.gnu.org/software/octave/

Note that I'm only showing links that have more than three occurrences. Anyone that wants to contribute can fork the repo and make a pull request!

Edit: I added duplicate detection, now the same link in an answer more than once only counts once. You can see the old, non-dupe-detecting, results in the revision history.

1
  • 2
    Cool! When SR becomes more popular than SO (maybe in 6 months?), this could be used to notice spam: if superclock.com has not been linked to in the past, but suddenly gets linked from 10 questions in a day, then it might be spam.
    – Nicolas Raoul Mod
    Apr 25, 2014 at 2:41

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .