# Sentiment Analysis of Chinese Product Reviews Based on Fusion of DUAL-Channel BiLSTM and Self-Attention

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Materials and Methods

#### 2.1. General Framework

#### 2.2. Specific Structure

#### 2.3. Experimental Data

## 3. Experiments and Results

#### 3.1. Computing Environment Configuration

#### 3.2. Evaluation Index

#### 3.3. Data Preprocessing

#### 3.4. Algorithm Parameter Selection

#### 3.5. Hyperparameter Setting

#### 3.6. Experimental Results

## 4. Discussion

#### 4.1. Comparison and Analysis with Other Methods

#### 4.2. Feature Analysis of the Proposed Dual-Channel Structure

#### 4.3. Shortcomings and Prospects

## 5. Conclusions

## Author Contributions

## Funding

## Data Availability Statement

## Conflicts of Interest

## References

- Punetha, N.; Jain, G. Bayesian game model based unsupervised sentiment analysis of product reviews. Expert Syst. Appl.
**2023**, 214, 119128. [Google Scholar] - Edara, D.C.; Vanukuri, L.P.; Sistla, V.; Kolli, V.K.K. Sentiment analysis and text categorization of cancer medical records with LSTM. J. Ambient. Intell. Humaniz. Comput.
**2023**, 14, 5309–5325. [Google Scholar] - Bhuvaneshwari, P.; Rao, A.N.; Robinson, Y.H.; Thippeswamy, M. Sentiment analysis for user reviews using Bi-LSTM self-attention based CNN model. Multimed. Tools Appl.
**2022**, 81, 12405–12419. [Google Scholar] - Luo, Y.; Yao, C.; Mo, Y.; Xie, B.; Yang, G.; Gui, H. A creative approach to understanding the hidden information within the business data using Deep Learning. Inf. Process. Manag.
**2021**, 58, 102615. [Google Scholar] - Liu, X.; Tang, T.; Ding, N. Social network sentiment classification method combined Chinese text syntax with graph convolutional neural network. Egypt. Inform. J.
**2022**, 23, 1–12. [Google Scholar] - Majumder, M.G.; Gupta, S.D.; Paul, J. Perceived usefulness of online customer reviews: A review mining approach using machine learning & exploratory data analysis. J. Bus. Res.
**2022**, 150, 147–164. [Google Scholar] - Jing, N.; Wu, Z.; Wang, H. A hybrid model integrating deep learning with investor sentiment analysis for stock price prediction. Expert Syst. Appl.
**2021**, 178, 115019. [Google Scholar] - Obiedat, R.; Qaddoura, R.; Ala’M, A.Z.; Al-Qaisi, L.; Harfoushi, O.; Alrefai, M.; Faris, H. Sentiment analysis of customers’ reviews using a hybrid evolutionary svm-based approach in an imbalanced data distribution. IEEE Access
**2022**, 10, 22260–22273. [Google Scholar] - Kewsuwun, N.; Kajornkasirat, S. A sentiment analysis model of agritech startup on Facebook comments using naive Bayes classifier. Int. J. Electr. Comput. Eng.
**2022**, 12, 2829–2838. [Google Scholar] - Dake, D.K.; Gyimah, E. Using sentiment analysis to evaluate qualitative students’ responses. Educ. Inf. Technol.
**2023**, 28, 4629–4647. [Google Scholar] - Benarafa, H.; Benkhalifa, M.; Akhloufi, M. WordNet Semantic Relations Based Enhancement of KNN Model for Implicit Aspect Identification in Sentiment Analysis. Int. J. Comput. Intell. Syst.
**2023**, 16, 3. [Google Scholar] [CrossRef] - Shamrat, F.M.J.M.; Chakraborty, S.; Imran, M.M.; Muna, J.N.; Billah, M.M.; Das, P.; Rahman, M.O. Sentiment analysis on twitter tweets about COVID-19 vaccines using NLP and supervised KNN classification algorithm. Indones. J. Electr. Eng. Comput. Sci.
**2021**, 23, 463–470. [Google Scholar] - Wawre, S.V.; Deshmukh, S.N. Sentiment classification using machine learning techniques. Int. J. Sci. Res.
**2016**, 5, 819–821. [Google Scholar] - Huq, M.R.; Ahmad, A.; Rahman, A. Sentiment analysis on Twitter data using KNN and SVM. Int. J. Adv. Comput. Sci. Appl.
**2017**, 8, 19–25. [Google Scholar] - Jiang, W.; Zhou, K.; Xiong, C.; Du, G.; Ou, C.; Zhang, J. KSCB: A novel unsupervised method for text sentiment analysis. Appl. Intell.
**2023**, 53, 301–311. [Google Scholar] - Bahdanau, D.; Cho, K.; Bengio, Y. Neural machine translation by jointly learning to align and translate. arXiv
**2014**, arXiv:1409.0473. [Google Scholar] - Vaswani, A.; Shazeer, N.; Parmar, N.; Uszkoreit, J.; Jones, L.; Gomez, A.N.; Kaiser, Ł.; Polosukhin, I. Attention is all you need. Adv. Neural Inf. Process. Syst.
**2017**, 30, 5999–6009. [Google Scholar] - Abdullah, T.; Ahmet, A. Deep learning in sentiment analysis: Recent architectures. Acm Comput. Surv.
**2022**, 55, 1–37. [Google Scholar] - Nguyen, H.D.; Huynh, T.; Hoang, S.N.; Pham, V.T.; Zelinka, I. Language-Oriented Sentiment Analysis Based on the Grammar Structure and Improved Self-attention Network. In Proceedings of the 15th International Conference, ENASE 2020, Prague, Czech Republic, 5–6 May 2020; pp. 339–346. [Google Scholar]
- Gan, C.; Wang, L.; Zhang, Z. Multi-entity sentiment analysis using self-attention based hierarchical dilated convolutional neural network. Future Gener. Comput. Syst.
**2020**, 112, 116–125. [Google Scholar] [CrossRef] - Yan, S.; Wang, J.; Song, Z. Microblog Sentiment Analysis Based on Dynamic Character-Level and Word-Level Features and Multi-Head Self-Attention Pooling. Future Internet
**2022**, 14, 234. [Google Scholar] [CrossRef] - Kudo, T.; Richardson, J. Sentencepiece: A simple and language independent subword tokenizer and detokenizer for neural text processing. arXiv
**2018**, arXiv:1808.06226. [Google Scholar] - Kim, Y. Convolutional neural networks for sentence classification. arXiv
**2014**, arXiv:1408.5882. [Google Scholar] - Kalchbrenner, N.; Grefenstette, E.; Blunsom, P. A convolutional neural network for modelling sentences. arXiv
**2014**, arXiv:1404.2188. [Google Scholar] - Zhou, C.; Sun, C.; Liu, Z.; Lau, F. A C-LSTM neural network for text classification. arXiv
**2015**, arXiv:1511.08630. [Google Scholar] - Ghourabi, A.; Mahmood, M.A.; Alzubi, Q.M. A Hybrid CNN-LSTM Model for SMS Spam Detection in Arabic and English Messages. Future Internet
**2020**, 12, 156. [Google Scholar] [CrossRef] - Gan, C.; Wang, L.; Zhang, Z.; Wang, Z. Sparse attention based separable dilated convolutional neural network for targeted sentiment analysis. Knowl.-Based Syst.
**2020**, 188, 104827. [Google Scholar] [CrossRef] - Liu, G.; Guo, J. Bidirectional LSTM with attention mechanism and convolutional layer for text classification. Neurocomputing
**2019**, 337, 325–338. [Google Scholar] [CrossRef] - Gan, C.; Feng, Q.; Zhang, Z. Scalable multi-channel dilated CNN–BiLSTM model with attention mechanism for Chinese textual sentiment analysis. Future Gener. Comput. Syst.
**2021**, 118, 297–309. [Google Scholar] [CrossRef] - Li, R.; Chen, H.; Feng, F.; Ma, Z.; Wang, X.; Hovy, E. Dual graph convolutional networks for aspect-based sentiment analysis. In Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 1: Long Papers), Virtual, 1–6 August 2021; pp. 6319–6329. [Google Scholar]
- Devlin, J.; Chang, M.; Lee, K.; Toutanova, K. BERT: Pre-Training of Deep Bidirectional Transformers for Language Understanding. arXiv
**2018**, arXiv:1810.04805. [Google Scholar] - Pipalia, K.; Bhadja, R.; Shukla, M. Comparative Analysis of Different Transformer Based Architectures Used in Sentiment Analysis. In Proceedings of the 2020 9th International Conference System Modeling and Advancement in Research Trends (SMART), Moradabad, India, 4–5 December 2020; pp. 411–415. [Google Scholar] [CrossRef]
- Beltagy, I.; Peters, M.E.; Cohan, A. Longformer: The Long-Document Transformer. arXiv
**2020**, arXiv:cs.CL/2004.05150. [Google Scholar]

**Figure 5.**Accuracy Variation Curve of Sac-BiLSTM on the Validation Set of the Online Shopping Review Dataset.

**Figure 6.**Accuracy Variation Curve of Sac-BiLSTM on the Validation Set of the Food Delivery Review Dataset.

**Figure 7.**Accuracy Variation Curve of Sac-BiLSTM on the Validation Set of the Weibo Comments Dataset.

**Figure 8.**Accuracy Comparison of Different Algorithms on the Test Set of the Online Shopping Review Dataset at Different Iterations.

Classes | Positive Comment Number | Negative Comment Number |
---|---|---|

Book | 2100 | 1751 |

Pad | 5000 | 5000 |

Phone | 1163 | 1158 |

Fruits | 5000 | 5000 |

Shampoo | 5000 | 5000 |

Water Heater | 100 | 475 |

Mengniu Dairy | 992 | 1041 |

Clothes | 5000 | 5000 |

Computer | 1996 | 1996 |

Hotel | 5000 | 5000 |

Experimental Environment | Environment Configuration |
---|---|

Operating System | Win10 |

CPU | i5-7300HQ CPU @ 2.50 GHz |

Memory | 8 GB |

Deep Learning Framework | TensorFlow 2.1.0-cpu |

Programming Language | Python 3.7 |

Word Segmentation Tool | jieba |

Feature Vector Training Tool | Word2Vec (gensim 3.8.3) |

Programming Environment | Anaconda 3 |

Real Class | Positive | Negative |
---|---|---|

Positive | TP (True Positive) | FN (False Negative) |

Negative | FP (False Positive) | TN (True Negative) |

Parameter | Character | Word |
---|---|---|

sg | Skip-gram | Skip-gram |

size | 300 | 300 |

min_count | 3 | 3 |

window | 10 | 10 |

workers | 4 | 4 |

Parameter | Value |
---|---|

BiLSTM units | 20 |

Convolution kernel size | 3, 5, 7 |

Number of convolution kernels | 128 |

Self_Attention Output dimension | 128 |

Dropout | 0.35 |

Batch_size | 64 |

Iterations | 15 |

Optimization Function | Adam |

**Table 6.**Text classification results of different algorithms on the Online Shopping Review dataset test set.

Algorithm | Feature | Acc | Precision | Recall | F1 |
---|---|---|---|---|---|

Static-CNN | Static Single-channel | 0.9188 | 0.9007 | 0.9390 | 0.9195 |

CNN-character | Static Single-channel | 0.9209 | 0.9006 | 0.9442 | 0.9219 |

S-Non-CNN | Non-static Dual-channel | 0.9336 | 0.9449 | 0.9190 | 0.9318 |

DCCNN | Non-static Dual-channel | 0.9372 | 0.9376 | 0.9352 | 0.9364 |

BiLSTM-CNN series | Non-static Single-channel | 0.9351 | 0.9408 | 0.9271 | 0.9339 |

BiLSTM-CNN parallel | Non-static Dual-channel | 0.9350 | 0.9413 | 0.9262 | 0.9337 |

Sac-BiLSTM | Non-static Dual-channel | 0.9463 | 0.9469 | 0.9409 | 0.9464 |

**Table 7.**Text classification results of different algorithms on the food delivery review dataset test set.

Algorithm | Feature | Acc | Precision | Recall | F1 |
---|---|---|---|---|---|

Static-CNN | Static Single-channel | 0.8376 | 0.8602 | 0.8432 | 0.8560 |

CNN-character | Static Single-channel | 0.8464 | 0.8457 | 0.8731 | 0.8592 |

S-Non-CNN | Non-static Dual-channel | 0.8494 | 0.8221 | 0.8930 | 0.8339 |

DCCNN | Non-static Dual-channel | 0.8564 | 0.8021 | 0.8477 | 0.8688 |

BiLSTM-CNN series | Non-static Single-channel | 0.8539 | 0.8057 | 0.8079 | 0.8538 |

BiLSTM-CNN parallel | Non-static Dual-channel | 0.8551 | 0.8341 | 0.8880 | 0.8602 |

Sac-BiLSTM | Non-static Dual-channel | 0.8776 | 0.8636 | 0.8980 | 0.8804 |

Algorithm | Feature | Acc | Precision | Recall | F1 |
---|---|---|---|---|---|

Static-CNN | Static Single-channel | 0.9261 | 0.9026 | 0.9492 | 0.9253 |

CNN-character | Static Single-channel | 0.9443 | 0.9389 | 0.9276 | 0.9130 |

S-Non-CNN | Non-static Dual-channel | 0.9508 | 0.8627 | 0.9494 | 0.9129 |

DCCNN | Non-static Dual-channel | 0.9566 | 0.9329 | 0.9587 | 0.9556 |

BiLSTM-CNN series | Non-static Single-channel | 0.9448 | 0.9332 | 0.9538 | 0.9434 |

BiLSTM-CNN parallel | Non-static Dual-channel | 0.9268 | 0.9031 | 0.9501 | 0.9260 |

Sac-BiLSTM | Non-static Dual-channel | 0.9775 | 0.9940 | 0.9592 | 0.9763 |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Yuan, Y.; Wang, W.; Wen, G.; Zheng, Z.; Zhuang, Z.
Sentiment Analysis of Chinese Product Reviews Based on Fusion of DUAL-Channel BiLSTM and Self-Attention. *Future Internet* **2023**, *15*, 364.
https://doi.org/10.3390/fi15110364

**AMA Style**

Yuan Y, Wang W, Wen G, Zheng Z, Zhuang Z.
Sentiment Analysis of Chinese Product Reviews Based on Fusion of DUAL-Channel BiLSTM and Self-Attention. *Future Internet*. 2023; 15(11):364.
https://doi.org/10.3390/fi15110364

**Chicago/Turabian Style**

Yuan, Ye, Wang Wang, Guangze Wen, Zikun Zheng, and Zhemin Zhuang.
2023. "Sentiment Analysis of Chinese Product Reviews Based on Fusion of DUAL-Channel BiLSTM and Self-Attention" *Future Internet* 15, no. 11: 364.
https://doi.org/10.3390/fi15110364