search

songyouwei / ABSA-PyTorch

Aspect Based Sentiment Analysis, PyTorch Implementations. 基于方面的情感分析，使用PyTorch实现。
MIT License
2.01k stars 524 forks source link

How to use for user input data? #74

Open abhinandansrivastava opened 5 years ago

abhinandansrivastava commented 5 years ago

Hi, Thanks for this repo.

I am trying to use ABSA, but after training how should I predict with the model,with User Input data..Also if you share the format it can be helpful.

Also whether this repo will help in finding Aspect terms or just the polarity of it.

Regards, Abhinandan

songyouwei commented 5 years ago
  1. For inference, please refer to infer_example.py.
  2. This repo focus on polarity classification task so far.
abhinandansrivastava commented 5 years ago

Thanks for reply.

I looked into infer_example.py.

Will this repo help us in find multi-aspect polarity term?

Regards, Abhinandan

songyouwei commented 5 years ago

You mean “aspect extraction”? No.

abhinandansrivastava commented 5 years ago

I meant, if in a sentence you have two aspects will this help in finding polarity for both aspect in a single sentence.

songyouwei commented 5 years ago

I'm not sure. I think the classification performance has nothing to do with the number of aspects in a sentence.

vivektop commented 3 years ago

Hi @songyouwei ,Thank you for this amazing work. I have the same doubt too. Ex : Service is good but food is bad',where aspects are: 'Service','food', Will the output of the model be like, Service : 1 , food : -1 ? Thanks