Synchronous Double-channel Recurrent Network for Aspect-Opinion Pair Extraction

Proceedings of the 58th Annual Meeting of the Association for Computational Linguistics, pages 6515–6524 July 5 - 10, 2020. c⃝2020 Association for Computational Linguistics 6515 Synchronous Double-channel Recurrent Network for Aspect-Opinion Pair Extraction Shaowei Chen, Jie Liu∗, Yu Wang, Wenzheng Zhang, Ziming Chi College of Artificial Intelligence, Nankai University, Tianjin, China {chenshaowei, yuwang17, wzzhang}@mail.nankai.edu.cn jliu@nankai.edu.cn zimingchi@163.com Abstract Opinion entity extraction is a fundamental task in fine-grained opinion mining. Related studies generally extract aspects and/or opin- ion expressions without recognizing the rela- tions between them. However, the relations are crucial for downstream tasks, including sentiment classification, opinion summariza- tion, etc. In this paper, we explore Aspect- Opinion Pair Extraction (AOPE) task, which aims at extracting aspects and opinion ex- pressions in pairs. To deal with this task, we propose Synchronous Double-channel Recurrent Network (SDRN) mainly consist- ing of an opinion entity extraction unit, a relation detection unit, and a synchroniza- tion unit. The opinion entity extraction unit and the relation detection unit are devel- oped as two channels to extract opinion enti- ties and relations simultaneously. Furthermore, within the synchronization unit, we design Entity Synchronization Mechanism (ESM) and Relation Synchronization Mechanism (RSM) to enhance the mutual benefit on the above two channels. To verify the perfor- mance of SDRN, we manually build three datasets based on SemEval 2014 and 2015 benchmarks. Extensive experiments demon- strate that SDRN achieves state-of-the-art per- formances. 1 Introduction Opinion entity extraction, which aims at identify- ing aspects and/or opinion expressions in review sentences, is an important task in fine-grained opin- ion mining. Recently, there have been considerable studies focused on this task. Specifically, Liu et al. (2012), Li and Lam (2017) and Li et al. (2018) ex- plored aspect term extraction, and Fan et al. (2019) extracted opinion phrases with given aspects. Mean- while, many studies dealt with aspect and opinion ∗Corresponding author. Review: The food was nice-looking and delicious. The result of Opinion Entity Extraction: Aspect: {food} Opinion Expression: {nice-looking, delicious} The result of Aspect-Opinion Pair Extraction: {food, nice-looking} {food, delicious} Figure 1: An example of task comparisons. The aspects and the opinion expressions are marked with red and blue, respectively. term co-extraction (Xu et al., 2013; Liu et al., 2015; Wang et al., 2017; Yu et al., 2019; Wang and Pan, 2019; Dai and Song, 2019). These studies have shown the importance of opinion entity extraction and achieved great progress. However, they neglect to recognize the relations between aspects and opin- ion expressions. While aspect-opinion relation detection is one of the key parts of an opinion mining system (Hu and Liu, 2004; Popescu and Etzioni, 2005; Zhuang et al., 2006), it is neglected or assumed given be- forehand, which leaves a significant gap to subse- quent opinion mining tasks. For instance, as shown in Figure 1, we can obtain the aspect {food} and the opinion expressions {nice-looking, delicious} from opinion entity extraction. Although both nice- looking and delicious express positive sentiment, they further describe food from the appearance and taste perspectives, respectively. Therefore, only …

Content truncated to 500 words