Inducing Target-Specific Latent Structures for Aspect Sentiment Classification

Proceedings of the 2020 Conference on Empirical Methods in Natural Language Processing, pages 5596–5607, November 16–20, 2020. c⃝2020 Association for Computational Linguistics 5596 Inducing Target-Specific Latent Structures for Aspect Sentiment Classification Chenhua Chen, Zhiyang Teng and Yue Zhang School of Engineering, Westlake University, China Institute of Advanced Technology, Westlake Institute for Advanced Study {chenchenhua, tengzhiyang}@westlake.edu.cn, yue.zhang@wias.org.cn Abstract Aspect-level sentiment analysis aims to rec- ognize the sentiment polarity of an aspect or a target in a comment. Recently, graph con- volutional networks based on linguistic de- pendency trees have been studied for this task. However, the dependency parsing ac- curacy of commercial product comments or tweets might be unsatisfactory. To tackle this problem, we associate linguistic depen- dency trees with automatically induced aspect- specific graphs. We propose gating mech- anisms to dynamically combine information from word dependency graphs and latent graphs which are learned by self-attention net- works. Our model can complement supervised syntactic features with latent semantic depen- dencies. Experimental results on five bench- marks show the effectiveness of our proposed latent models, giving significantly better re- sults than models without using latent graphs. 1 Introduction Aspect-level sentiment analysis aims to classify the sentiment polarities towards specific aspect terms in a given sentence (Jiang et al., 2011; Dong et al., 2014; Vo and Zhang, 2015). Aspects are also called opinion targets, which can typically be product or service features in customer reviews. For example, in the user comment “The environment is roman- tic, but the food is horrible”, the sentiments of the two aspects “environment” and “food” are posi- tive and negative, respectively. The main challenge of aspect-level sentiment analysis is to effectively model the interaction between the aspect and its surrounding contexts. For example, identifying that “romantic” instead of “horrible” as the opinion word is the key to correctly classifying the senti- ment of “environment”. Recently, graph convolutional networks (GCNs; Kipf and Welling (2017)) over dependency i complained to the manager , but he was not even apologetic (a) An example dependency tree from Stanford CoreNLP parser2. the portions are small but being that the food was so good makes up for that . (b) A latent graph for the aspect term “portion”. the portions are small but being that the food was so good makes up for that . (c) A latent graph for the aspect term “food”. trees (Marcheggiani and Titov, 2017; Zhang et al., 2019; Sun et al., 2019b; Wang et al., 2020) have re- ceived much research attention. It has been shown to be more effective for learning aspect-specific representations than traditional sentence encoders without considering graph structures (Tang et al., 2016a,b; Liu and Zhang, 2017; Li et al., 2018a). Intuitively, dependency trees allow a model to bet- ter represent the correlation between aspect terms and their relevant opinion words. However, the existing methods suffer from two potential limi- tations. First, dependency parsing accuracies can be relatively low on noisy texts such as tweets, blogs and review comments, which are the main sources of aspect-level …

Content truncated to 500 words