Skip to content

A Position-aware Bidirectional Attention Network for Aspect-level Sentiment Analysis, PyTorch implementation.

License

Notifications You must be signed in to change notification settings

hiyouga/PBAN-PyTorch

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

5 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

PBAN-PyTorch

PyTorch implementation of Gu et al.'s COLING 2018 work.

LICENSE

Requirement

Dataset

Based on the restaurant and laptop dataset of SemEval-2014 Task 4.

Restaurant Dataset

Polarity #Positive #Negative #Neutral
Train 2164 807 637
Test 728 196 196

Laptop Dataset

Polarity #Positive #Negative #Neutral
Train 994 870 464
Test 341 128 169

Usage

Train the model:

python train.py --model_name pban --dataset restaurant

Show help message and exit:

python train.py -h

Implemented models

LSTM

Tang, Duyu, et al. "Effective LSTMs for Target-Dependent Sentiment Classification." Proceedings of COLING 2016, the 26th International Conference on Computational Linguistics: Technical Papers. 2016. [pdf]

lstm

ATAE-LSTM

Wang, Yequan, Minlie Huang, and Li Zhao. "Attention-based LSTM for aspect-level sentiment classification." Proceedings of the 2016 conference on empirical methods in natural language processing. 2016. [pdf]

atae_lstm

PBAN

Gu, Shuqin, et al. "A Position-aware Bidirectional Attention Network for Aspect-level Sentiment Analysis." Proceedings of the 27th International Conference on Computational Linguistics. 2018. [pdf]

pban

Performance

Restaurant Dataset

Three-class

Model In Paper This Code
LSTM 74.28 77.68
ATAE-LSTM 77.20 78.30
PBAN 81.16 80.89

Two-class

Model In Paper This Code
LSTM - -
ATAE-LSTM 90.90 90.26
PBAN 91.67 92.32

Laptop Dataset

Three-class

Model In Paper This Code
LSTM 66.45 71.00
ATAE-LSTM 68.70 71.32
PBAN 74.12 74.76

Two-class

Model In Paper This Code
LSTM - -
ATAE-LSTM 87.60 87.63
PBAN 87.81 87.42

Acknowledgements

  • Some of the code is borrowed from songyouwei.
  • Using this code means you have read and accepted the copyrights set by the dataset providers.

License

MIT

About

A Position-aware Bidirectional Attention Network for Aspect-level Sentiment Analysis, PyTorch implementation.

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages