I work with big catalog data, and with lots of human input. We have these internal classifications but people mis-assign items all the time.
I was horrified when I learned about the state of our datasets. The only reliable features are the ones that are exposed to the customers on the website, because when they're broken people can actually see them.
I was horrified when I learned about the state of our datasets. The only reliable features are the ones that are exposed to the customers on the website, because when they're broken people can actually see them.