|
|
|
|
|
by prepend
3029 days ago
|
|
Zagat likely had much more data than what was publicly available. I have no visibility, but can imagine non-public data useful for training like individual reviewer biases, reviews through their edit lifecycle, stuff like that. |
|