Sentiment Analysis of Short Texts Based on Parallel DenseNet

Yan, Luqi; Han, Jin; Yue, Yishi; Zhang, Liu; Qian, Yannan

doi:10.32604/cmc.2021.016920

Cited by 5 publications

(4 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Zhang et al [18] proposed a character-level convolutional neural network, which changed the input mode at the word level and divided sentences into single English letters, numbers, and other characters as input, alleviating the problems caused by segmentation errors and spelling errors. Yan et al [19] added two parallel DenseNet Networks to the CNN model to pay attention to the global features of the text, and combined the extracted different features to realize the sentiment analysis of the short text. Gan [20] proposed a Separable Extended Convolutional Neural Network (SA-SDCCN) based on sparse attention, which is mainly composed of a separable convolutional module, sparse attention layer, and output layer.…”

Section: Deep Learning Methodsmentioning

confidence: 99%

A Short Text Sentiment Analysis Model Combined with Attention Mechanism of Bidirectional Affective Words

Xu,

Li,

Wang

et al. 2023

JCEIM

View full text Add to dashboard Cite

Short text sentiment in social media platforms has important research value. Due to the convolution kernel parameter sharing and pooling operation, convolutional neural network training speed is fast, and the effect is also good. However, most work such as max-pooling, average-pooling, more or less discard some of the secondary features, which makes the final expression of emotion possibly biased. So, in this paper, the attention mechanism applies to the pooling layer of CNN, and a Convolutional Neural Network based on double sentiment word attention pooling (DSA-CNN) is proposed. Each feature has its weight to be calculated. We also note that in the attention mechanism of text classification, the attention query vector is usually randomly initialized during network training, rather than using an existing vector--the semantic information of the previous moment, as in machine translation. Therefore, we use the emotion dictionary and emotion corpus to train bidirectional affective word vectors, so that the emotion-related features can interact with them in the attention. The experiment shows that DSA-CNN has achieved better performance than the classical classification model in a dataset of Weibo nCoV Data, NLPCC 2014, and yf_dianping. The accuracy of the model is 2.18% higher than that of the second-best model. Besides, the convergence rate of DSA-CNN is also significantly improved. The highest accuracy was achieved only in the third epochs, while CNN using the max-pooling needed 14 epochs.

show abstract

Section: Deep Learning Methodsmentioning

confidence: 99%

A Short Text Sentiment Analysis Model Combined with Attention Mechanism of Bidirectional Affective Words

Xu,

Li,

Wang

et al. 2023

JCEIM

View full text Add to dashboard Cite

show abstract

“…Without relying solely on stacking convolutional layers, a CNN with dense connections was proposed by Wang et al [6] to reuse existing multi-scale sentiment features and flexibly generate larger-scale features. Yan et al [12] used a feature extraction block based on a convolution operation and a feature extraction block with dense connections as its feature extraction module, and their parallelism saved training time and reduced training iterations.…”

Section: Cnns and Resnets In Sa Tasksmentioning

confidence: 99%

“…Currently, there are two approaches to alleviate the above limitations: convolution filters with various window sizes in a layer and densely connected layers [6,11,12]. The first approach utilizes filters with different window sizes to extract multi-scale sentiment features.…”

Section: Introductionmentioning

confidence: 99%

Position-Wise Gated Res2Net-Based Convolutional Network with Selective Fusing for Sentiment Analysis

Zhou

Zeng

Zou

et al. 2023

Entropy

View full text Add to dashboard Cite

Sentiment analysis (SA) is an important task in natural language processing in which convolutional neural networks (CNNs) have been successfully applied. However, most existing CNNs can only extract predefined, fixed-scale sentiment features and cannot synthesize flexible, multi-scale sentiment features. Moreover, these models’ convolutional and pooling layers gradually lose local detailed information. In this study, a new CNN model based on residual network technology and attention mechanisms is proposed. This model exploits more abundant multi-scale sentiment features and addresses the loss of locally detailed information to enhance the accuracy of sentiment classification. It is primarily composed of a position-wise gated Res2Net (PG-Res2Net) module and a selective fusing module. The PG-Res2Net module can adaptively learn multi-scale sentiment features over a large range using multi-way convolution, residual-like connections, and position-wise gates. The selective fusing module is developed to fully reuse and selectively fuse these features for prediction. The proposed model was evaluated using five baseline datasets. The experimental results demonstrate that the proposed model surpassed the other models in performance. In the best case, the model outperforms the other models by up to 1.2%. Ablation studies and visualizations further revealed the model’s ability to extract and fuse multi-scale sentiment features.

show abstract

“…By analysing the microblog texts, they explored the degree and importance of the mutual influence between the two emotions. Yan et al [17] proposed a parallel DenseNet to realise sentiment analysis of short texts based on traditional densely connected convolutional networks (DenseNet). The model can learn the local and global features of short text at the same time, and it shows better performance on six different databases than other basic models.…”

Section: Literature Reviewmentioning

confidence: 99%

DRMM: A novel data mining-based emotion transfer detecting method for emotion prediction of social media

Shi

Xue

Yin

et al. 2022

Journal of Information Science

View full text Add to dashboard Cite

With the progress of the Internet and information technology, emotion analysis has been applied to analyse the emotional orientation and evolution trend of online public opinion of online tweets. At present, most of the existing methods use econometric model and machine learning algorithm to predict the trend of online public opinion. Although these methods have achieved good prediction results, they do not take into account the influence of internal factors on network public opinion prediction, such as mutual migration among emotion classes. The emotion may change dynamically because different events trigger it in the evolution process. In this view, this article proposes a novel method, called Deviation Rule Markov Model (DRMM), to predict the emotional change trend of Internet users in online public opinion by analysing the correlation between Internet users’ emotional categories. Structurally, the proposed DRMM involves various processes such as pre-processing, emotion classification, data mining and transfer prediction. For the processing of network comment data, the proposed model initially undergoes pre-processing to delete unnecessary data. Then, the extended fuzzy emotion ontology is used to annotate the emotion class of the comment data. Besides, an extended association rule mining algorithm is used in the emotion association analysis process to obtain the transfer probability between emotion classes. Moreover, Markov chain is used to construct an emotional state transition matrix to predict the transition probability of positive or negative emotions. According to the predicted single emotion transfer probability results, the analytic hierarchy process is used to assign values to different emotion classes, and finally, the transfer probability of the overall emotion in a certain period is obtained. Compared with the actual case, the mean absolute error (MAE) and root mean square error (RMSE) of the proposed model are 2.7119 and 3.7254, respectively, which has good prediction performance.

show abstract

Sentiment Analysis of Short Texts Based on Parallel DenseNet

Cited by 5 publications

References 7 publications

A Short Text Sentiment Analysis Model Combined with Attention Mechanism of Bidirectional Affective Words

A Short Text Sentiment Analysis Model Combined with Attention Mechanism of Bidirectional Affective Words

Position-Wise Gated Res2Net-Based Convolutional Network with Selective Fusing for Sentiment Analysis

DRMM: A novel data mining-based emotion transfer detecting method for emotion prediction of social media

Contact Info

Product

Resources

About