search

yangheng95 / LCF-ATEPC

codes for paper A Multi-task Learning Model for Chinese-oriented Aspect Polarity Classification and Aspect Term Extraction
MIT License
191 stars 45 forks source link

Regard Dataset #43

Open Astudnew opened 2 years ago

Astudnew commented 2 years ago

Hello in the restaurant dataset, I find 4 tags for B-ASP and I-ASP I understand that 0 is negative, -1 is natural, 2 is positive what about 1 such as Food B-ASP 1? and why camera dataset has just 3 labels 0 , -1, and 2? to reformat my dataset (semeval 2016) in another language what should I follow?

Astudnew commented 2 years ago

Also, I found in the training file for the restaurant dataset you consider just 3 labels (0.1.2), I misunderstand? "test_f1 = f1_score(torch.argmax(test_apc_logits_all, -1).cpu(), test_polarities_all.cpu(), labels=[0, 1, 2], average='macro')"