Comparison of Deep Learning Sentiment Analysis Methods, Including LSTM and Machine Learning

Purpose of research. The purpose of the study is to evaluate certain machine learning models in data processing based on speed and efficiency related to the analysis of sentiment or consumer opinions in business intelligence. To highlight the existing developments, an overview of modern methods and models of sentiment analysis is given, demonstrating their advantages and disadvantages.

Materials and methods. In order to improve the semester analysis process, organized using existing methods and models, it is necessary to adjust it in accordance with the growing changes in information flows today. In this case, it is crucial for researchers to explore the possibilities of updating certain tools, either to combine them or to develop them to adapt them to modern tasks in order to provide a clearer understanding of the results of their treatment. We present a comparison of several deep learning models, including convolutional neural networks, recurrent neural networks, and long-term and shortterm bidirectional memory, evaluated using different approaches to word integration, including Bidirectional Encoder Representations from Transformers (BERT) and its variants, FastText and Word2Vec. Data augmentation was conducted using a simple data augmentation approach. This project uses natural language processing (NLP), deep learning, and models such as LSTM, CNN, SVM TF-IDF, Adaboost, Naive Bayes, and then combinations of models.

The results of the study allowed us to obtain and verify model results with user reviews and compare model accuracy to see which model had the highest accuracy results from the models and their combination of CNN with LSTM model, but SVM with TF-IDF vectoring was most effective for this unbalanced data set. In the constructed model, the result was the following indexes: ROC AUC - 0.82, precision - 0.92, F1 - 0.82, Precision - 0.82, and Recall - 0.82. More research and model implementation can be done to find a better model.

Conclusion. Natural language text analysis has advanced quite a bit in recent years, and it is possible that such problems will be completely solved in the near future. Several different models in ML and CNN with the LSTM model, but SVM with the TF-IDF vectorizer proved most effective for this unbalanced data set. In general, both deep classification algorithm. A combination of both approaches can also learning and feature-based selection methods can be used to solve be used to further improve the efficiency of the algorithm. some of the most pressing problems. Deep learning is useful when the most relevant features are not known in advance, while feature-based

1. Romanov A.S., Kurtukova A.V., Sobolev A.A. Determination of the age of the author of the text based on deep neural network models. Information. 2020; 11(12): 589.

2. Shlomo A. E., Mosher K., Galit A. Text classification by style: what newspaper do I read? In the collection. From the AAAI Workshop on Text Categorization; 1998: 1-4.

3. Bay S., Kolter Dzh.Z, Koltun V. Empirical evaluation of general convolutional and recurrent networks for sequence modeling. Preprint arXiv arXiv. 2018; 2: 1803-01271.

4. Konno A., Shvenk KH., Barro L. et. al. Very deep convolutional networks for text classification. Preprint arXiv arXiv. 2017; 2: 1606-01781.

5. Zhang KH., Chzhao J., Lekun Y. Symbol-level convolutional networks for text classification. Preprint arXiv arXiv. 2016; 3: 1509-01626.

6. In' U., K. Kannan K. et. al. Comparative study of CNN and RNN for natural language processing. Preprint arXiv arXiv. 2017; 1: 1702.

7. Yogatama D., Dayyer Chr., Ling U. et. al. Generative and discriminative text classification using recurrent neural networks. Preprint arXiv arXiv. 2017; 2: 1703-01898.

8. Balakrishnan V., Lok P.YA., Rakhim KH.A. A semi-managed approach to detecting sentiment and emotion based on surveys of digital payments.J Supercomput. 2021; 77: 3795-3810.

9. Karosiya A.E., Koel'o G.P., Sil'va A.E. Investment Strategies Applied to the Brazilian Stock Market: A Methodology Based on Sentiment Analysis Using Deep Learning. Expert Syst Application. 2021: 184.

10. TSzin N., Vu Z., Vang KH. A hybrid model integrating deep learning with investor sentiment analysis for stock price prediction. Expert Syst Application. 2021: 178.

11. Yadav A., Dzha K.K., Sharan A. et. al. Analysis of sentiment in financial news using an unsupervised approach. Proced Comput Sci. 2020; 167: 589-598.

12. Chzhan YU., Khan R., TSze M. et. al. Social media analytics platform for improving operations and service management: A study of the retail pharmacy industry. Change Prediction Technology in Soc. 2021: 163.

13. Vu Dzh.Dzh., Chang S.T. Exploring Consumer Sentiment for Online Retail Services: A Thematic Approach. J Retail Consumer. 2020; 55: 102145.

14. Chzhan Dzh., Chzhan A., Lyu D. et. al. Extracting consumer preference for air purifiers based on detailed sentiment analysis of online reviews. Knowledge Based System. 2021: 228.

15. Syuy F., Pan Z., Sya R. E-commerce Product Review and Sentiment Classification Based on Naive Bayesian Continuous Learning. Process Management Inf. 2020: 6(57).

16. Tapariya A, Bagla T. Sentiment Analysis: Predicting Product Review Scores Using Online Customer Reviews. 2020. DOI: 10.2139/ssrn.3655308.

17. Kolon-Ruis S., Segura-Bedmar I. Comparison of deep learning architectures for sentiment analysis in drug reviews. J Biomed Inform. 2020: 110.

18. Vu F., Shi Z., Dong Z. et. al. SenBERT-CNN Based Online Product Review Sentiment Analysis. International Conference on Machine Learning and Cybernetics (ICMLC). 2020: 229-234.

19. Pota M., Ventura M., Katelli R. et. al. Efficient BERT-based pipeline for Twitter sentiment analysis: a case study in Italian. Sensors. 2021; 21(1): 133.

20. Shorten K., Khoshgoftaar T. M., Furkht B. Text data extension for deep learning. Big Data. 2021; 8: 101.

21. Krizhevskiy A., Sutskever I., Khinton G.Ye Imagenet classification using deep convolutional neural networks. Commun ACM. 2017: 84–90.

22. Kobayashi S. Contextual Augmentation: Incrementing Data with Words with Paradigmatic Relationships. V NAACL HLT. 2018; 2: 452-457.

23. Duong KH.T., Nguyen-Tkhi T.A. Review: preprocessing methods and data augmentation for sentiment analysis. Computational Network. 2021; 8: 1.

24. Chzhou S., Chen K., Van KH. Active deep learning method for user-controlled mood classification. Neurocomputing. 120: 536-546.

25. Den L., Khinton G., Kingsberi B. New Types of Deep Learning Neural Networks for Speech Recognition and Related Applications: A Review. IEEE Int. Conf. Acoustics. Speech signal processing. 2013: 859-860.

26. Bengio S., Deng L., Laroshel' KH., Salakhutdinov R.I. Introduction by Guest Editors: A Special Section on the Study of Deep Architectures. IEEE Trans Pattern Anal Mach Intell. 2013; 35(8): 1795-1797.

27. Arnol'd L., Rebekki S., Sheval'ye S. et. al. Introduction to deep learning. Esann. 2011: 479-488.

28. Go Y., Lyu YU., Erlemans A. et. al. Deep learning for visual understanding: a review. Neurocomputing.2016; 187: 27-48.

29. Guan' Z. Yan Dzh. Restrained self-learning: a semi-supervised sentiment classification method for Chinese microblogging. Proceedings of the 6th International Joint Conference on Natural Language Processing. 2013: 455-462.

30. Chen Z., Mukerdzhi A., Lyu B. Aspect extraction with automated prior knowledge learning. In ACL Proceedings. 2014: 347-358.

31. Prakash V. Dzh., Nit'ya D. L. A review of semi-supervised learning methods. International Journal of Computer Trends and Technologies. 2014; 8(1): 25-29.

32. Guidance on sentiment analysis [Internet]. Available from: https://monkeylearn.com/sentiment-analysis/.

33. Basic guide to sentiment analysis [Internet]. Available from: https://www.telusinternational.com/insights/ai-data/article/the-essential-guide-tosentiment-analysis.