Identify ABSA single- or compound-tasks to try to implement
Based on the comprehensive and recent overview provided by Zhang et al. (2022), several methods were systematically identified for further investigation:
Potential tasks of interest:
- ATE (aspect term extraction)
- ASC (aspect sentiment classification)
- E2E (end 2 end, ATE + ASC)
- ASTE (aspect sentiment triple extraction, ATE, OTE, ASC)
Either a pipeline method consisting of an ATE and ASC methods, or one of the compound methods may suffice for ABSA on tweet data.
All 4 above mentioned potential tasks were investigated concerning their documented performance on well-known datasets (mostly SemEval 2014, 2015, 2016 but also Mitchell et al.'s 2013 twitter dataset). The top scoring methods for each task are selected for more in-depth inspection. Priority was also given to methods that scored particularly well on the twitter dataset.
Top methods per task are listed below.
ATE Methods
- Q. Wang, Z. Wen, Q. Zhao, M. Yang, and R. Xu, “Progressive self-training with discriminator for aspect term extraction,” in EMNLP, 2021, pp. 257–268 [method 33]
- Chen and T. Qian, “Enhancing aspect term extraction with soft prototypes,” in EMNLP, 2020, pp. 2107–2117. [method 32]
- Yang, K. Li, X. Quan, W. Shen, and Q. Su, “Constituency lattice encoding for aspect term extraction,” in COLING, 2020, pp. 844–855. [method 29-bert]
ASC Methods
- C. Sun, L. Huang, and X. Qiu, “Utilizing BERT for aspect-based sentiment analysis via constructing auxiliary sentence,” in NAACL-HLT, 2019, pp. 380–385. [method 127]
- J. Cheng, S. Zhao, J. Zhang, I. King, X. Zhang, and H. Wang, “Aspect-level sentiment classification with HEAT (hierarchical attention) network,” in ACMCIKM, 2017, pp. 97–106. [method 57]
- B. Wang, T. Shen, G. Long, T. Zhou, and Y. Chang, “Eliminating sentiment bias for aspect-level sentiment classification with unsupervised opinion extraction,” in Findings of EMNLP, 2021, pp. 3002–3012. [method 73]
End 2 End Methods
- G. Yu, J. Li, L. Luo, Y. Meng, X. Ao, and Q. He, “Self question-answering: Aspect-based sentiment analysis by role flipped machine reading comprehension,” in Findings of EMNLP, 2021, pp. 1331–1342. [method 87]
- Y. Liang, F. Meng, J. Zhang, Y. Chen, J. Xu, and J. Zhou, “An iterative multi-knowledge transfer network for aspect-based sentiment analysis,” in Findings of EMNLP, 2021, pp. 1768–1780. [method 86]
- H. Luo, L. Ji, T. Li, D. Jiang, and N. Duan, “GRACE: Gradient harmonized and cascaded labeling for aspect-based sentiment analysis,” in Findings of EMNLP, 2020, pp. 54–64. [method 85]
ASTE Methods
- Z. Wu, C. Ying, F. Zhao, Z. Fan, X. Dai, and R. Xia, “Grid tagging scheme for aspect-oriented fine-grained opinion extraction,” in Findings of EMNLP, 2020, pp. 2576–2585. [method 75]
- L. Xu, Y. K. Chia, and L. Bing, “Learning span-level interactions for aspect sentiment triplet extraction,” in ACL-IJCNLP, 2021, pp. 4755–4766. [method 97]
- W. Zhang, X. Li, Y. Deng, L. Bing, and W. Lam, “Towards generative aspect-based sentiment analysis,” in ACL-IJCNLP, 2021, pp. 504–510 [method 95]