Proceedings of the 2019 Conference on Empirical Methods in Natural Language Processing
and the 9th International Joint Conference on Natural Language Processing, pages 6280–6285,
Hong Kong, China, November 3–7, 2019.
c
2019 Association for Computational Linguistics
6280
A Challenge Dataset and Effective Models for Aspect-Based Sentiment
Analysis
Qingnan Jiang
1
, Lei Chen
1
, Ruifeng Xu
2,3
, Xiang Ao
4
, Min Yang
1∗
1
Shenzhen Institutes of Advanced Technology, Chinese Academy of Sciences
2
Department of Computer Science, Harbin Institute of Technology (Shenzhen)
3
Peng Cheng Laboratory
4
Institute of Computing Technology, Chinese Academy of Sciences
jqnthomask@gmail.com, lei.chen@siat.ac.cn, xuruifeng@hit.edu.cn
aoxiang@ict.ac.cn, min.yang@siat.ac.cn
Abstract
Aspect-based sentiment analysis (ABSA) has
attracted increasing attention recently due to
its broad applications. In existing ABSA
datasets, most sentences contain only one as-
pect or multiple aspects with the same senti-
ment polarity, which makes ABSA task degen-
erate to sentence-level sentiment analysis. In
this paper, we present a new large-scale Multi-
Aspect Multi-Sentiment (MAMS) dataset, in
which each sentence contains at least two dif-
ferent aspects with different sentiment polar-
ities. The release of this dataset would push
forward the research in this field. In addi-
tion, we propose simple yet effective CapsNet
and CapsNet-BERT models which combine
the strengths of recent NLP advances. Ex-
periments on our new dataset show that the
proposed model significantly outperforms the
state-of-the-art baseline methods
1
.
1 Introduction
Aspect-based sentiment analysis (ABSA) aims at
identifying the sentiment polarity towards the spe-
cific aspect in a sentence. An target aspect refers
to a word or a phrase describing an aspect of an
entity. For example, in the sentence “The decor
is not special at all but their amazing food makes
up for it”, there are two aspect terms “decor” and
“food”, and they are associated with negative and
positive sentiment respectively.
Recently, neural network methods have domi-
nated the study of ABSA since these methods can
be trained end-to-end and automatically learn im-
portant features. (Wang et al., 2016) proposed to
learn an embedding vector for each aspect, and
these aspect embeddings were used to calculate
the attention weights to capture important infor-
mation with regard to the given aspects. (Tang
∗
Min Yang is corresponding author
1
Data and code can be found as: https://github.com/siat-
nlp/MAMS-for-ABSA
et al., 2016b) developed the deep memory network
to compute the importance degree and text repre-
sentation of each context word with multiple at-
tention layers. (Ma et al., 2017) introduced the in-
teractive attention networks (IAN) to interactively
learn attentions in contexts and targets, and gen-
erated the representations for target and context
words separately. (Xue and Li, 2018) proposed to
extract sentiment features with convolutional neu-
ral networks and selectively output aspect related
features for classification with gating mechanisms.
Subsequently, Transformer (Vaswani et al., 2017)
and BERT based methods (Devlin et al., 2018)
have shown high potentials on ABSA task. There
are also several studies attempting to simulate the
process of human reading cognition to further im-
prove the performance of ABSA (Lei et al., 2019;
Yang et al., 2019).
So far, several ABSA datasets have been con-
structed, including SemEval-2014 Restaurant Re-
view dataset, Laptop Review dataset (Pontiki
et al., 2014) and Twitter dataset (Dong et al.,
2014). Although these three datasets have since
become the benchmark datasets for the ABSA
task, most sentences in these datasets consist
of only one aspect or multiple aspects with the
same sentiment polarity (see Table 1)
2
, which
makes aspect-based sentiment analysis degener-
ate to sentence-level sentiment analysis. Based on
our empirical observation, the sentence-level sen-
timent classifiers without considering aspects can
still achieve competitive results with many recent
ABSA methods (see TextCNN and LSTM in Ta-
ble 3). On the other hand, even advanced ABSA
methods trained on these datasets can hardly dis-
tinguish the sentiment polarities towards different
aspects in the sentences that contain multiple as-
pects and multiple sentiments.
2
ATSA and ACSA represent aspect-term and aspect-
category sentiment analysis, respectively.