|
A Framework for Urban Crisis Monitoring and Warning Based on Persian Data in Social Networks
|
| |
|
|
|
Abstract: (7 Views) |
Early detection of urban crises is one of the fundamental needs of urban management, which can play a significant role in reducing damages. This research proposes a framework based on analyzing Persian tweets on the social network X (formerly Twitter) for the early detection of urban crises. In this method, the messages related to the crises are first screened using weak labeling. Then, a semi-supervised classification model is trained using a limited amount of manually labeled data to classify new messages . Finally, using the model-predicted labeled data and time series analysis, bursts in the volume of crisis-related messages are identified, and early warnings are generated. Evaluation of the model on real data (more than 300,000 tweets) demonstrates the acceptable performance of the proposed method. Furthermore, the temporal analysis successfully detected bursts patterns aligned with real-world events. This approach not only facilitates the development of a system relying on unstructured social network data for the early detection of natural and service-related urban crises but also it can be extended to predict the early signs of social crises. |
|
| |
|
|
Full-Text [PDF 1398 kb]
(5 Downloads)
|
Type of Study: Research Article |
Subject:
Soft Security and Cognitive Threads
Received: 2026/05/14 | Accepted: 2026/09/5 | Published: 2026/09/15
|
|
|
|
|
|
|
| Add your comments about this article |
|
|
|
