CRIPAC-DIG / TextING

[ACL 2020] Tensorflow implementation for "Every Document Owns Its Structure: Inductive Text Classification via Graph Neural Networks"
180 stars 57 forks source link

[Issue] Suspicion Regarding Experimental Results and Code Accuracy #27

Open Lucienxhh opened 10 months ago

Lucienxhh commented 10 months ago

In approximately 3 out of 10 runs, there is a chance that positive and negative labels will be swapped, meaning that positive samples are treated as negative samples and vice versa. The experimental results reported in the paper are questionable, and it seems that the root cause may be an error in this script for one-hot encoding.