A Novel Counterfactual Data Augmentation Method for Aspect-Based Sentiment Analysis

Dongming Wu, Lulu Wen, Chao Chen, Zhaoshu Shi
Proceedings of the 15th Asian Conference on Machine Learning, PMLR 222:1479-1493, 2024.

Abstract

Aspect-based-sentiment-analysis (ABSA) is a fine-grained sentiment evaluation task, which analyzes the emotional polarity of the evaluation aspects. Generally, the emotional polarity of an aspect exists in the corresponding opinion expression, whose diversity has great impact on model’s performance. To mitigate this problem, we propose a novel and simple counterfactual data augmentation method to generate opinion expressions with reversed sentiment polarity. In particular, the integrated gradients are calculated to locate and mask the opinion expression. Then, a prompt combined with the reverse expression polarity is added to the original text, and a Pre-trained language model (PLM), T5, is finally was employed to predict the masks. The experimental results shows the proposed counterfactual data augmentation method performs better than current augmentation methods on three ABSA datasets, i.e. Laptop, Restaurant, and MAMS.

Cite this Paper

BibTeX
@InProceedings{pmlr-v222-wu24a, title = {A Novel Counterfactual Data Augmentation Method for Aspect-Based Sentiment Analysis}, author = {Wu, Dongming and Wen, Lulu and Chen, Chao and Shi, Zhaoshu}, booktitle = {Proceedings of the 15th Asian Conference on Machine Learning}, pages = {1479--1493}, year = {2024}, editor = {Yanıkoğlu, Berrin and Buntine, Wray}, volume = {222}, series = {Proceedings of Machine Learning Research}, month = {11--14 Nov}, publisher = {PMLR}, pdf = {https://proceedings.mlr.press/v222/wu24a/wu24a.pdf}, url = {https://proceedings.mlr.press/v222/wu24a.html}, abstract = {Aspect-based-sentiment-analysis (ABSA) is a fine-grained sentiment evaluation task, which analyzes the emotional polarity of the evaluation aspects. Generally, the emotional polarity of an aspect exists in the corresponding opinion expression, whose diversity has great impact on model’s performance. To mitigate this problem, we propose a novel and simple counterfactual data augmentation method to generate opinion expressions with reversed sentiment polarity. In particular, the integrated gradients are calculated to locate and mask the opinion expression. Then, a prompt combined with the reverse expression polarity is added to the original text, and a Pre-trained language model (PLM), T5, is finally was employed to predict the masks. The experimental results shows the proposed counterfactual data augmentation method performs better than current augmentation methods on three ABSA datasets, i.e. Laptop, Restaurant, and MAMS.} }
Endnote
%0 Conference Paper %T A Novel Counterfactual Data Augmentation Method for Aspect-Based Sentiment Analysis %A Dongming Wu %A Lulu Wen %A Chao Chen %A Zhaoshu Shi %B Proceedings of the 15th Asian Conference on Machine Learning %C Proceedings of Machine Learning Research %D 2024 %E Berrin Yanıkoğlu %E Wray Buntine %F pmlr-v222-wu24a %I PMLR %P 1479--1493 %U https://proceedings.mlr.press/v222/wu24a.html %V 222 %X Aspect-based-sentiment-analysis (ABSA) is a fine-grained sentiment evaluation task, which analyzes the emotional polarity of the evaluation aspects. Generally, the emotional polarity of an aspect exists in the corresponding opinion expression, whose diversity has great impact on model’s performance. To mitigate this problem, we propose a novel and simple counterfactual data augmentation method to generate opinion expressions with reversed sentiment polarity. In particular, the integrated gradients are calculated to locate and mask the opinion expression. Then, a prompt combined with the reverse expression polarity is added to the original text, and a Pre-trained language model (PLM), T5, is finally was employed to predict the masks. The experimental results shows the proposed counterfactual data augmentation method performs better than current augmentation methods on three ABSA datasets, i.e. Laptop, Restaurant, and MAMS.
APA
Wu, D., Wen, L., Chen, C. & Shi, Z.. (2024). A Novel Counterfactual Data Augmentation Method for Aspect-Based Sentiment Analysis. Proceedings of the 15th Asian Conference on Machine Learning, in Proceedings of Machine Learning Research 222:1479-1493 Available from https://proceedings.mlr.press/v222/wu24a.html.

Related Material