From [[Main Library - Chatterbot]]
The Ubuntu training corpus is an enormous (3gb) collection of conversational text from Ubuntus tech-support system.
It’s heavily biased and hopelessly garbage but should allow for a decent starting point.
Rendering context...