4 ms·

I am amazed at the amount of data that had to be collected for this to correctly guess 5 for 5 of my characters. some of them were kind of obscure too. I wonder

by dc2k08 18y ago

I am amazed at the amount of data that had to be collected for this to correctly guess 5 for 5 of my characters. some of them were kind of obscure too. I wonder if they didn't crowd-source it some how. edit: with the first incorrect guess of seven, i get it. they are crowd-sourcing the data every time it loses. brilliant.

liuliu 18y ago
A thought, can we use wikipedia info to boost up the accuracy of 20q? For that we can have a huge near accurate data at the beginning. Considering the paper on CIKM 2008 "Learning to Link with Wikipedia", I think that it is practical.
- paulgb 18y ago
  Or freebase.