Article: Managing digital public opinion: a case study on developing a deep learning monitoring system for Weibo Journal: International Journal of Information and Communication Technology (IJICT) 2026 Vol.27 No.52 pp.1 - 27 Abstract: Government agencies struggle to track and respond to public sentiment on social media platforms like Weibo. This case study describes the design and development of a monitoring system for an anonymous municipal government in China, leveraging deep learning to analyse sentiment and emerging topics. The case details the system architecture, implementation challenges, and how the outputs can be used for targeted public communication. To achieve effective management of social public opinion, this article uses deep learning and clustering algorithms to process public opinion information on the Weibo platform and establishes a Weibo public opinion analysis system. Focusing on user blog posts and comments, we first use distributed crawlers to obtain data, and then complete preprocessing through cleaning and word segmentation. Emotion analysis is implemented to obtain sentiment polarity and probability, and to explore potential themes using a latent Dirichlet allocation topic model. The experimental results show that the established model has high accuracy in emotion classification. Using real Weibo data, the emotional value change curve of netizens is plotted to determine the impact of topics on netizens' emotions. The system supports targeted public opinion intervention for governmental use. Inderscience Publishers - linking academia, business and industry through research

Title: Managing digital public opinion: a case study on developing a deep learning monitoring system for Weibo

Authors: Lin Zhu; Li Zhuang

Addresses: School of Electronics and Computer Engineering, Southeast University Chengxian College, Nanjing, 210088, China ' School of Electronics and Computer Engineering, Southeast University Chengxian College, Nanjing, 210088, China

Abstract: Government agencies struggle to track and respond to public sentiment on social media platforms like Weibo. This case study describes the design and development of a monitoring system for an anonymous municipal government in China, leveraging deep learning to analyse sentiment and emerging topics. The case details the system architecture, implementation challenges, and how the outputs can be used for targeted public communication. To achieve effective management of social public opinion, this article uses deep learning and clustering algorithms to process public opinion information on the Weibo platform and establishes a Weibo public opinion analysis system. Focusing on user blog posts and comments, we first use distributed crawlers to obtain data, and then complete preprocessing through cleaning and word segmentation. Emotion analysis is implemented to obtain sentiment polarity and probability, and to explore potential themes using a latent Dirichlet allocation topic model. The experimental results show that the established model has high accuracy in emotion classification. Using real Weibo data, the emotional value change curve of netizens is plotted to determine the impact of topics on netizens' emotions. The system supports targeted public opinion intervention for governmental use.

Keywords: Weibo; public opinion; analysis.

DOI: 10.1504/IJICT.2026.153708

International Journal of Information and Communication Technology, 2026 Vol.27 No.52, pp.1 - 27

Received: 18 Jul 2025
Accepted: 21 Oct 2025
Published online: 21 May 2026 *

Title: Managing digital public opinion: a case study on developing a deep learning monitoring system for Weibo

Keep up-to-date