Multi-head self-attention based gated graph convolutional networks for aspect-based sentiment classification
- PDF / 2,865,788 Bytes
- 20 Pages / 439.642 x 666.49 pts Page_size
- 37 Downloads / 201 Views
Multi-head self-attention based gated graph convolutional networks for aspect-based sentiment classification Luwei Xiao1 · Xiaohui Hu1 · Yinong Chen2 · Yun Xue1 · Bingliang Chen1 · Donghong Gu1 · Bixia Tang1 Received: 29 March 2020 / Revised: 30 July 2020 / Accepted: 19 October 2020 / © Springer Science+Business Media, LLC, part of Springer Nature 2020
Abstract Aspect-based sentiment classification aims to predict the sentiment polarity of specific aspects appeared in a sentence. Nowadays, most current methods mainly focus on the semantic information by exploiting traditional attention mechanisms combined with recurrent neural networks to capture the interaction between the contexts and the targets. However, these models did not consider the importance of the relevant syntactical constraints. In this paper, we propose to employ a novel gated graph convolutional networks on the dependency tree to encode syntactical information, and we design a Syntax-aware Context Dynamic Weighted layer to guide our model to pay more attention to the local syntax-aware context. Moreover, Multi-head Attention is utilized for capturing both semantic information and interactive information between semantics and syntax. We conducted experiments on five datasets and the results demonstrate the effectiveness of the proposed model. Keywords Aspect-based sentiment classification · Multi-head Self-Attention · Gated graph convolutional networks · Syntax-aware Context Dynamic Weighted
1 Introduction Opinion mining and sentiment analysis [1] are tasks which focus on identifying the sentence-level or document-level sentiment polarities. Different from predicting the over Xiaohui Hu
[email protected] Luwei Xiao [email protected] 1
Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials, GPETR Center for Quantum Precision Measurement, SPTE, School of Physics and Telecommunication Engineering, South China Normal University, Guangzhou 510006, China
2
School of Computing, Informatics, and Decision Systems Engineering, Arizona State University, Tempe, AZ 85287, USA
Multimedia Tools and Applications
all sentiment polarity of a sentance, as a fine-grained task in natural language processing (NLP), aspect-based sentiment classification (ABSC) [24] aims at predicting sentiments (e.g., positive, negative or neutral) from sentance about a given aspect. For instance, “Battery life could be better but overall the price is higher!”, the emotional polarity of “Battery life” is negative and as for “price” is positive. Since traditional sentence-level sentiment classification methods are not able to identify polarities for specific aspects in a fine-grained level, this task is capacity of obtaining valuable information which is helpful for people to assist them in making decisions in their life (e.g.,shopping, travel, etc). Aspect-level sentiment classification is a promising research topics in NLP. The core ides of traditional approaches is to extract and exploit some hand-making features, for example, sentimental lexicon an
Data Loading...
