Nessvectors for San Diego Python User Group
I had a lot of fun playing with words at the monthly Python User group meeting in San Diego this week. Congratulations to Torin Panick @torrinp for winning a free copy of NLP in Action. For those of you that missed out, I’ll give out one free eBook code and a 42% discount code next time. And I’ll be a bit more organized about the competition ;).
If you’re wondering, a “nessvector” isn’t a real NLP thing. That’s just what I call an explainable word vector that has your own custom “dimensions”. The examples we used during the meetup had dimensions for “peopleness”, “animalness”, “placeness”, and “femaleness”:
>>> nessvector('cat') placeness -1.560885 peopleness -0.216057 animalness 2.906886 conceptness -1.442183 femaleness 0.312240
Lots of animalness femaleness (femininity) and not any placeness at all.
What do you think a “dog” nessvector will look like? More or less femininity than a cat?
>>> nessvector('dog') placeness -1.453316 peopleness -0.589015 animalness 3.112393 conceptness -1.225242 femaleness 0.155180
What about “herb”?
>>> nessvector('herb') placeness -0.340492 peopleness 0.036387 animalness 0.647465 conceptness -0.356363 femaleness 0.013003
And for the 20 questions game, what about “tiger”?
>>> nessvector('tiger') placeness -1.283883 peopleness -0.140355 animalness 2.350045 conceptness -1.342111 femaleness 0.416304