An error in 'read' function about TNet. Specifically in utils.py

[tangjialong]

Hello, THX for your shared code of TNet. However, I observed that an error in "read" function about TNet. Specifically, in utils.py: words.append(t.strip(end)) target_words.append(t.strip(end))

if use t.strip(), it will cause an error , such as 'nicki/n'.strip('/n') the ouptut is 'icki' rather than 'nicki' when I try to use t[:-2] to replace t.strip(): words.append(t[:-2]) target_words.append(t[:-2])

I find that the best accuracy and F1-score reported in your paper can not be achieved on those three datasets (Laptop, REST, TWITTER).

[lixin4ever]

Yes, you are right. Such pre-processing error indeed affects the final performances (1-2% accuracy drop). You can refer to issue 4 for more information.

Considering this pre-processing error and the issue that theano is no longer maintained, I highly recommend you to use ABSA-pytorch, a pytorch-based implementation of many ABSA models including TNet, for reproduction.