rockerest’s avatarrockerest’s Twitter Archive—№ 16,748

    1. …in reply to @schizanon
      @yesiwebto :P Well then I'll have to suggest affinity. When I was building a text search utility, having match affinity (based on stemming, in my case, but you may not need stems) would have unlocked a ton of fuzzy matching options.
  1. …in reply to @rockerest
    @yesiwebto In your case, you'd want to compute an affinity of one file name to another file name. Dashes wouldn't lower the affinity much, but entirely unique words would be a big hit.
    1. …in reply to @rockerest
      @yesiwebto Then you could set some threshold like ">95% affinity deduped automatically, 75-95% they're displayed as unknown, <75% they're considered different" OR SOMETHING, I dunno man I'm just riffing