Aspect-based Sentiment Analysis with Type-aware Graph Convolutional Networks and Layer Ensemble

Proceedings of the 2021 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies, pages 2910–2922 June 6–11, 2021. ©2021 Association for Computational Linguistics 2910 Aspect-based Sentiment Analysis with Type-aware Graph Convolutional Networks and Layer Ensemble Yuanhe Tian♥⇤, Guimin Chen~⇤, Yan Song♠~† ♥University of Washington ~Shenzhen Research Institute of Big Data ♠The Chinese University of Hong Kong (Shenzhen) ♥yhtian@uw.edu ~chenguimin@foxmail.com ♠songyan@cuhk.edu.cn Abstract It is popular that neural graph-based models are applied in existing aspect-based sentiment analysis (ABSA) studies for utilizing word re- lations through dependency parses to facilitate the task with better semantic guidance for an- alyzing context and aspect words. However, most of these studies only leverage depen- dency relations without considering their de- pendency types, and are limited in lacking effi- cient mechanisms to distinguish the important relations as well as learn from different layers of graph based models. To address such limita- tions, in this paper, we propose an approach to explicitly utilize dependency types for ABSA with type-aware graph convolutional networks (T-GCN), where attention is used in T-GCN to distinguish different edges (relations) in the graph and attentive layer ensemble is proposed to comprehensively learn from different layers of T-GCN. The validity and effectiveness of our approach are demonstrated in the exper- imental results, where state-of-the-art perfor- mance is achieved on six English benchmark datasets. Further experiments are conducted to analyze the contributions of each component in our approach and illustrate how different layers in T-GCN help ABSA with quantitative and qualitative analysis.1 1 Introduction Aspect-based sentiment analysis (ABSA) processes fine-grained sentiment polarities towards specific aspects, where in many cases, it is required to iden- tify different sentiments for multiple aspects in the same context. For example, in the sentence “The drink menu is limited but the wines are excellent.”, the sentiment polarity towards “drink menu” is neg- ative while that towards “wines” is positive; an *Equal contribution. †Corresponding author. 1The code and models involved in this paper are released at https://github.com/cuhksz-nlp/ASA-TGCN. Figure 1: An example sentence (including the aspect term “drink menu”) with its dependency parsing result. ABSA system may predict wrong if it fails to cap- ture the important contextual information for each aspects. Therefore, to model such contextual infor- mation, neural models (e.g., Bi-LSTM and Trans- former (Vaswani et al., 2017)) have been widely used for ABSA and demonstrated to be useful for this task (Wang et al., 2016; Tang et al., 2016a; Chen et al., 2017; Ma et al., 2017; Fan et al., 2018). As a further enhancement of encoding contextual information for ABSA, there are studies (Sun et al., 2019; Huang and Carley, 2019; Zhang et al., 2019a) using graph convolutional networks (GCN) to learn from a graph that is often built over the dependency parsing results of the input texts. As a result, the GCN models are able to learn from distant word- word relations that are more helpful to ABSA. How- ever, GCN models used in these studies are limited by omitting the information …

Content truncated to 500 words