7

@dfox
Is it possible you added an auto tag feature that pucks up technical words and tags them?

Comments
  • 1
    @dfox - nice idea
  • 1
    @Jumpshot44 think about it. Just make a huge list of tags and do this in python:

    for tag in taglist:
    for word in rant:
    if tag in word:
    ##ADD TAG CODE HERE
  • 0
    OH MY FUCKING GOD.

    I SPELT PICK AS PUCK.

    May almighty god @dfox forgive me amen
  • 0
    @Pgdevpatryk Relax, you can edit rants now
  • 0
    @Neat Nope 5 mins after you can"t
  • 0
    A repost-check for pictures would also be nice
  • 0
  • 0
    @Hazen the problem would be it would have to be recent. Because you cant really scroll down to last months post to see if the picture was used. The algorithm would need to google-level good because of white text that sometimes overlays "meme" pictures.
  • 1
    @Pgdevpatryk
    The memes aren't the problem I guess, because if it's another text on it it's not a repost.
    I know an image board with ~ 1.5 million posts which has an open-source community made repost-check that isn't perfect but not bad either. But I didn't look into it so I can't tell how it works. Anyways this is it:
    https://github.com/ReneHollander/...
  • 2
    @Hazen Also we need Do not Disturb mode.
  • 1
    #puck
  • 4
    If you search for a keyword using search or by tapping a tag term, it does a full text search, so tags are more for supplemental info. Pulling out words from the rant content and adding them to tags would be redundant, though could make sense for longer rants, where keywords get lost in paragraphs, but most rants are fairly short.

    Image comparison is something we've looked into, some good libraries out there that could get us most of the way there, but lots of meme images get used with different text to a very different effect, so more likely we'll start with OCR on the images to auto mark the same text as reposts
  • 0
    @Pgdevpatryk let me speed that code up for you:
    * tag list is always sorted
    * sort words in rant
    * linear pass through both lists simultaneously to find matches

    there - it's in n log n now and not just in quadratic :)
Add Comment