没有合适的资源?快使用搜索试试~ 我知道了~
A Text Sentiment Classification Modeling Method Based on Coordin...
1 下载量 35 浏览量
2021-02-08
13:48:20
上传
评论
收藏 1.94MB PDF 举报
温馨提示
The major challenge that text sentiment classification modeling faces is how to capture the intrinsic semantic, emotional dependence information and the key part of the emotional expression of text. To solve this problem, we proposed a Coordinated CNNLSTM-Attention(CCLA) model. We learned the vector representations of sentence with CCLA unit. Semantic and emotional information of sentences and their relations are adaptively encoded to vector representations of document. We used softmax regressio
资源推荐
资源详情
资源评论
Chinese Journal of Electronics
Vol.28, No.1, Jan. 2019
A Text Sentiment Classification Modeling
Method Based on Coordinated
CNN-LSTM-Attention Model
∗
ZHANG Yangsen
1,2
, ZHENG Jia
1
, JIANG Yuru
1,2
, HUANG Gaijuan
1,2
and CHEN Ruoyu
1,2
(1. Institute of Intelligent Information Processing, Beijing Information Science and Technology University,
Beijing 100192, China)
(2. Beijing Laboratory of National Economic Security Early-Warning Engineering,Beijing 100192,China)
Abstract — The major challenge that text sentiment
classification modeling faces is how to capture the
intrinsic semantic, emotional dependence information and
the key part of the emotional expression of text. To
solve this problem, we proposed a Coordinated CNN-
LSTM-Attention(CCLA) model. We learned the vector
representations of sentence with CCLA unit. Semantic and
emotional information of sentences and their relations are
adaptively encoded to vector representations of document.
We used softmax regression classifier to identify the
sentiment tendencies in the text. Compared with other
methods, the CCLA model can well capture the local
and long distance semantic and emotional information.
Experimental results demonstrated the effectiveness of
CCLA model. It shows superior performances over several
state-of-the-art baseline methods.
Key words — Coordinated CNN-LSTM-Attention,
Sentiment analysis, Text modeling, Semantic information.
I. Introduction
Text sentiment classification modeling is a funda-
mental problem in the field of Nature language processing
(NLP) and is a crux to understand user intention
in product reviews or social networks
[1,2]
. The core
of text sentiment classification modeling is to capture
semantic features from variable-length text units. As a
traditional method, the bag-of-words model
[3]
is the most
common and popular vector representations method for
texts because of its efficiency, simplicity and surprising
accuracy. But the bag-of-words model treats sentence or
document as an unordered collection of words. Lacking
word order, different sentences can have the exactly same
representation, given that the same words are used.
Until now, some machine learning algorithms have
achieved good results on text sentiment classification
modeling
[4]
, but with the deep learning models have
achieved remarkable effects in the field of speech
recognition and computer vision in recent years, order-
sensitive models based on the neural networks model such
as Recursive neural networks (RNNs), Recurrent neural
networks (RNN), Convolutional neural networks (CNN),
Long short-term memory (LSTM) and attention model
are becoming increasingly popular due to their ability
to capture word order information and further learn
the semantic and emotional information from text. Deep
learning comes from traditional neural network models.
It is not just a multi-layer network but emphasizes the
extraction of hidden features and higher-level abstract
features.
II. Related Work
1. Deep learning model
RNNs have been proved effective in modeling text
semantics
[5−7]
. However, it need to construct semantic
tree and its performance depends on the accuracy of the
semantic tree. But, the semantic relationship between
two sentences may not be able to form a tree structure.
RNN do not need to build the semantic tree
[8]
and it
can capture the context information over long distances.
However, RNN is a bias model, or to be more specific, a
positive model, in which the relatively backward words
in the text occupy a more dominant position. At the
same time, RNN also have the problem of exploding and
vanishing gradient.
In order to solve the semantic bias problem of RNN,
it is proposed to use CNN for text semantic modeling.
∗
Manuscript Received Aug. 4, 2017; Accepted May 29, 2018. This work is supported by the National Natural Science
Foundation of China (No.61772081, No.61602044) and the Science and Technology Development Project of Beijing Municipal Education
Commission(No.KM201711232014).
© 2019 Chinese Institute of Electronics. DOI:10.1049/cje.2018.11.004
A Text Sentiment Classification Modeling Method Based on Coordinated CNN-LSTM-Attention Model 121
CNN is always used in conjunction with the pooling
technology to find the most useful information from
the text. However, the existing CNN models always
use relatively simple convolutional kernels, such as fixed
windows
[9]
. It brings great limitations to the semantic
representation of text.
Recently, RNN with LSTM units have revitalized
become a popular architecture due to its representa-
tional ability and effectiveness at capturing long-term
dependence information
[10,11]
. LSTM can also solve the
problem of exploding and vanishing gradient of RNN.
A LSTM block may be described as an “intelligent”
network unit that can remember a value which contain
some history information for an arbitrary length of time.
The attention model is also a popular language
model in recent years. Bahdanau et al.
[12]
think that the
traditional Encoder-Decoder model generates fixed-length
intermediate semantics representation is the bottleneck
to enhance the performance of neural machine translation
model. To solve this problem, they proposed an automatic
alignment model between the output and the source
sentence and achieved good performance. This alignment
model is the basis of most of the present attention models.
The attention model is good at capturing the key parts
that are important to the semantic representation.
2. Deep learning model for sentiment classifi-
cation
Deep learning has been proved effective for text sen-
timent classification tasks
[13,14]
. For the text sentiment
classification task, the magic of deep learning is that
it can learn continuous and multidimensional semantic
and emotional representations with different grains. The
deep learning model for text sentiment classification
modeling always includes two steps: the first step is
learning distributed vector representations of words, next
step is utilizing the principle of compositionality which
states the meaning of a longer expression (e.g. sentence or
document) depends on the meanings of its constituents,
to construct the vector representation of sentences and
documents based on the vector representations of the
words.
For learning distributed vector representations of
words, Maas
[15]
presented a model that uses a mix
of unsupervised and supervised techniques to learn
the vector representations of word, which can capture
semantic information as well as rich emotional content.
Mikolov
[13,16]
proposed two novel deep learning models
for computing continuous vector representations of words
from larger corpora and built a well-known open source
project dubbed word2vec. Pennington
[17]
introduced the
GloVe model, which is an unsupervised global log-bilinear
regression model for learning the representations of word.
In order to construct the representation of sentences
and documents, inspired by learning vector representa-
tions of words, Le
[18]
proposed the paragraph vector
model, an unsupervised learning algorithm that learns
fixed-length feature representations of sentences. Kalch-
brenner
[14]
designed a dynamic convolutional architecture
dubbed the Dynamic convolutional neural network
(DCNN) that uses dynamic k-max pooling, which is
a global pooling operation over linear sequences to
capture the most useful information from the text. Tai
[10]
generalized LSTM to Tree-LSTM architecture where
each LSTM unit combines information from its children
units to improve semantic representation for sentences.
For some specific domain text sentiment classification
modeling, deep learning model also has a good effect.
Zhang
[19]
built a language model based on RNN to handle
negative sentences and double negative sentences for
Chinese Microblog sentiment classification.
Text modeling should take full account of the
semantic relation of text in short distance and the
semantic dependence of text in long distance, as well as
the key part of the emotional expression. In order to solve
the defects of various single deep learning models for text
sentiment modeling, we proposed a new model dubbed
Coordinated CNN-LSTM-attention(CCLA) model which
can encode the semantic and emotional information of
text for sentiment classification task as shown in Fig.1.
Fig. 1. The architecture of the CCLA model
III. Model
Before introducing our approach for sentence model-
ing, we will introduce word vector representation model.
In the process of word modeling, each word is represented
as a low dimensional, continuous and real-valued vector
and associated with a point in a vector space
[20]
. Then
all the word vectors are mapped into a word embedding
matrix R
d×|V|
, where d is the dimension of word vectors
and |V| is the vocabulary size of embedding matrix.
1. Sentence vector representation model
To construct vector representations for sentences,
CNN and LSTM are two state-of-the-art models. In this
paper, we combined the advantages of CNN and LSTM
and incorporated with the attention model to construct
the CCLA unit to encode the semantic and emotional
information for sentences. The architecture of the CCLA
unit is given in Fig. 2.
剩余6页未读,继续阅读
资源评论
weixin_38688097
- 粉丝: 5
- 资源: 928
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
最新资源
- 职工上、下班交通费补贴规定.docx
- 房地产公司圣诞活动策划方案.docx
- 全球旅游与经济指标数据集,旅游影响因素数据集,旅游与收入数据(六千六百多条数据)
- 公司下午茶费用预算.xlsx
- 下午茶.docx
- 毕设和企业适用springboot计算机视觉平台类及在线平台源码+论文+视频.zip
- 2014年度体检项目.xls
- 年度员工体检项目.xls
- 年度体检.xlsx
- 毕设和企业适用springboot跨境电商平台类及虚拟现实体验平台源码+论文+视频.zip
- 毕设和企业适用springboot平台对接类及全球电商管理平台源码+论文+视频.zip
- 数据库-sqlite客户端-sqlite-访问sqlite数据库
- 住宅小区汽车超速检测及报警系统设计(单片机源码+图+报告)
- 毕设和企业适用springboot区块链技术类及客户关系管理平台源码+论文+视频.zip
- 毕设和企业适用springboot区块链技术类及音频处理平台源码+论文+视频.zip
- 毕设和企业适用springboot区块链交易平台类及交通信息平台源码+论文+视频.zip
资源上传下载、课程学习等过程中有任何疑问或建议,欢迎提出宝贵意见哦~我们会及时处理!
点击此处反馈
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功