News for hackers. Stuff that matters.
Re:Show HN: Sense2vec model trained on all 2015 Reddit comments by Houshalter on Monday February 15, 2016 @09:46PM (#11107600)
Not necessarily a bug on their end, those urls are actually from a bot, AutoWikibot. Which uses the word "NSFW" in every post, so all of it's formatting is highly correlated with it: https://www.reddit.com/user/AutoWikibot

The last line (Parent commenter can toggle NSFW or delete...) has this formatting:

    ^Parent ^commenter ^can [^toggle ^NSFW](/message/compose?to=autowikibot&subject=AutoWikibot NSFW toggle&message=%2Btoggle-nsfw+ct3omf8) ^or[](#or) [^delete](/message/compose?to=autowikibot&subject=AutoWikibot Deletion&message=%2Bdelete+ct3omf8)^. ^Will ^also ^delete ^on ^comment ^score ^of ^-1 ^or ^less. ^| [^(FAQs)](/r/autowikibot/wiki/index) ^| [^Mods](/r/autowikibot/comments/1x013o/for_moderators_switches_commands_and_css/) ^| [^Call ^Me](/r/autowikibot/comments/1ux484/ask_wikibot/)
You can see how that formatting might fuck up their nice natural language processor and tokenizer.

There are no comments on this comment yet.