19 Questions

Think of something a common person would know about, then I ask you questions and guess what it is.



19 Questions is a machine-learning game that finds out about things by playing games with users. The algorithm is based on Bayesian statistics, compared to other similar games use neural networks.

See also: notes on the the Bayesian/entropy approach and the GitHub project.


Questions in knowledge base:564
Objects in knowledge base:4,073
Connections in knowledge base:231,619
Games played:16,574
New user-submitted objects:228
Total object entropy:11.7