News for hackers. Stuff that matters.
Re:Organizing My Emails with a Neural Net by jorangreef on Sunday January 24, 2016 @03:11AM (#10961910)
Orthogonal sparse bigrams are another interesting tokenization method, the authors of this paper tested it out against CRM114 with good results:

http://www.siefkes.net/ie/winnow-spam.pdf

Re:Organizing My Emails with a Neural Net by Lanzaa on Monday January 25, 2016 @02:17PM (#10969214)
Orthogonal sparse bigrams (OSBs) have been implemented in CRM114 since the publication of that paper.