Forums

SemEval 2018 Task 10 Capturing Discriminative Attributes Forum

Go back to competition Back to thread list Post in this thread

> Questions about the dataset

In the dataset description, we saw that the training data is "automatically generated", hence "noisy". Does that mean the labels in the training set are completely random? Or is it generated according to some criteria?

Posted by: Tetheras @ March 21, 2018, 2:33 p.m.
Post in this thread