---

Journal of Vcibration Testing and System Dynamics 4(2) (2020) 163--171 | DOI:10.5890/JVTSD.2020.06.003

Yang Xiang, Hong Song

School of Automation and Information Engineering, Sichuan University of Science and Engineering, Zigong, Sichuan 643000, China

Download Full Text PDF

 

Abstract

Aiming at the problems that traditional machine learning algorithms such as SVM rely too much on emotional dictionary, grammatical mechanism and low recognition rate in processing text sentiment by natural language, a different ratio CNN-ATTBiLSTM algorithm based on the algorithm is proposed in this paper. First, the CNN algorithm is used to process the local features, and the CNN feature fusion method is used to integrate ATTBiLSTM (two-way long-term and short-term memory network based on attention mechanism) at the output, Then the two algorithms are combined in different ratios to find the highest F1 value as the combination ratio. Finally, the algorithm proposed in this paper is compared with the traditional machine learning algorithms and the proposed algorithms in the experimental data set. The results show that the new algorithm is 0.85% better than the equal weight ratio CNN-ATTBiLSTM algorithm. So it is the best algorithm compared with other algorithms.

References

1. [1]	Zhao, W., Qin B., and Liu T. (2010), Text sentiment analysis, Journal of Software, 21(8), 1834-1848.

2. [2]	Kim, Y. (2014), Convolutional neural networks for sentence classification, Belgium, 5, 87-93.

3. [3]	Hochreiter, S. and Schmidhuber, J. (1997), Long short-term memory, Neural Computation, 9, 1735-1780.

4. [4]	Zhang, Y. and Xia, H.B. (2018), Sentiment Analysis Method Based on Weakly Supervised Pre-trained CNN Model, Computer Engineering and Applications, 54(13), 27-33.

5. [5]	Tu, M.Z., Yan, Y., and Yong, H. (2017), Sentiment Analysis of Weibo Events Based on CNN-SVM and Forwarding Tree, Information Engineering, 3(03), 77-85.

6. [6]	Jian, F.F., Li, M.D., Tao, W.G., and Chen, L. (2019), Twitter sentiment analysis based on multi-channel LSTM-CNN model, Journal of Langfang Teachers College(Natural Science Edition), 19(02), 21-24+37.

7. [7]	Guo, H., Xu, W., Lu, K., and Tang, Q. (2019), A Short Text Similarity Calculation Method Based on CNN and BiLSTM , Information Technology and Cybersecurity, 38(06), 61-64+68.

8. [8]	Gao, Y.L., Zhan, X.G., and Chi, C.Y. (2018), Sentiment Analysis Based on CNN-LSTM Model, Journal of Liaoning University of Science and Technology, 41(06), 469-474.

9. [9]	Liu, N. and Ren, F.J. (2019), Emotion classification using a CNN LSTM-based model for smooth emotional synchronization of the humanoid robot REN-XIN, PloSone,14(5), 80-85.

10. [10]	Kalchbrenner, N., Grefenstette, E., and Blunsom, P. (2014), A convolutional neural network for modelling sentences, Proceedings of the 2014 Annual Meeting of the Association for Computational Linguistics. Balti- more: ACL, 27, 655-665.

11. [11]	Kai, S.R. and Christopher, D.M. (2015), Improved Semantic Representations From Tree-Structured Long Short-Term Memory Networks Beijing, China, Association for Computational Linguistics, 33, 1556-1566.

12. [12]	Fan, G., Shi, S.W., Xin, Q.L., Hong, F.C., and Xiao, L.C. (2019), Research on Text Mining of Railway Safety Supervisors Performance Based on BiLSTM and CRF, Journal of Physics, Conference Series, 12(5), 63-69.

13. [13]	Sun, L. (2018), A Method of Chinese Named Entity Recognition Based on CNN-BILSTM-CRF Model, ICYCSEE Steering Committee, 38, 596-603.

14. [14]	Zhang, Y. and Wallace, B.C. (2017), A Sensitivity Analysis of (and Practitioners' Guide to) Convolutional Neural Networks for Sentence Classification, Taipei, China, International Joint Conference on Natural Lan- guage Processing, 55, 253-263.

15. [15]	Apeksha, S.D. and Nyavanandi, S., and Ludwig, A. (2019), Performance Evaluation of Deep Neural Networks Applied to Speech Recognition: RNN, LSTM and GRU, Journal of Artificial Intelligence and Soft Computing Research, 9(4), 91-96.