Re:30% of Google's Emotions Dataset Is Mislabeled
by mdorazio on Thursday July 14, 2022 @10:59AM
(#32096872)
> I wonder how others do this kind of thing. Assuming I have 100k text blurbs I want to classify of sentiment and/or emotion and want at least 95% accuracy, what are the options?
Fwiw, in a similar situation I did about half myself and farmed out the other half to my retired parents. Using trusted people was the only way I found I could get high accuracy without spending thousands of dollars. But, as I frequently tell people, the hard part of ML isn’t the model/code, it’s the training data.
Fwiw, in a similar situation I did about half myself and farmed out the other half to my retired parents. Using trusted people was the only way I found I could get high accuracy without spending thousands of dollars. But, as I frequently tell people, the hard part of ML isn’t the model/code, it’s the training data.