one of the tables in the article mentions something called "word2tensor" but google doesn't throw up anything
except this tweet which seems to be from a conference https://twitter.com/ic/status/756918600846356480?lang=en
does anyone have any info about it?