Authors : Fanrong Meng; Zhan Wen; Lanyun Chen; Rui Jiang; Dehao Ren

Volume/Issue : Volume 10 - 2025, Issue 2 - February

Google Scholar : https://tinyurl.com/242pryn7

Scribd : https://tinyurl.com/yc5b5yvp

DOI : https://doi.org/10.5281/zenodo.14979414

Abstract : In the post-epidemic era, public opinion monitoring and early warning for COVID-19 prevention and control have become vital research areas. This study aims to explore key technologies to enhance the capability of monitoring and early warning for public opinion regarding epidemic prevention and control. First, the study focuses on the collection and processing of public opinion data. By constructing a large-scale data collection system that integrates social media, news platforms, forums, and other channels, this study enables the real-time acquisition of public opinion information related to the epidemic. At the same time, natural language processing and text mining techniques are employed to clean, classify, and analyze the sentiment of large-scale text data, facilitating the extraction of valuable insights. Second, to enhance public opinion monitoring, this study introduces an emotion classification model based on deep learning. The model is compared with traditional machine learning approaches to evaluate its effectiveness in distinguishing the emotional tone of public opinion texts and analyzing individuals' attitudes and emotional responses toward the epidemic. To optimize performance, an early stopping mechanism is implemented during training to prevent overfitting, halting the process when validation loss ceases to improve after a specified number of iterations. Additionally, hyperparameter optimization is conducted using a grid search, systematically exploring various parameter combinations to identify the optimal configuration. Data balance is carefully maintained to enhance the model’s predictive accuracy and robustness, ensuring reliable and highquality results.

Keywords : Public Opinion Monitoring, Emotion Classification, Deep Learning, Hyperparameter Optimization.

References :

1. Xie Erhu. Research on Key Technologies of Data Preprocessing in Data Mining [J]. Science and Technology Bulletin (in Chinese), 2013,29 (12): 211-213. DOI: 10.13774/j.cnki. kjtb.2013.12.066
2. Zhuohao W, Dong W, Qing L I. Keyword Extraction from Scientific Research Projects Based on SRP‐TF‐IDF[J]. Chinese Journal of Electronics, 2021, 30(4): 652-657.
3. Sah S, Surendiran B, Dhanalakshmi R, et al. Covid‐19 cases prediction using SARIMAX Model by tuning hyperparameter through grid search cross‐validation approach[J]. Expert Systems, 2023, 40(5): e13086.
4. Je J, Ma Z, Lei J, et al. Advanced dropout: A model-free methodology for bayesian dropout optimization[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2021, 44(9): 4605-4625.
5. Gygi, J. P., Kleinstein, S. H. and Guan, L. (2023) ‘Predictive overfitting in immunological applications: Pitfalls and solutions’, Human Vaccines & Immunotherapeutics, 19(2). doi: 10.1080/21645515.2023.2251830.
6. Li X, Chang D, Tian T, et al. Large-margin regularized softmax cross-entropy loss[J]. IEEE access, 2019, 7: 19572-19578.
7. Bigoni C, Cadic-Melchior A, Morishita T, et al. Optimization of phase prediction for brain-state dependent stimulation: a grid-search approach[J]. Journal of Neural Engineering, 2023, 20(1): 016039.