Research on Evolution Model and Crisis Prevention and Control of New Media Public Opinion

Kongyu Yang, Ruijie Miao2017 · DOI

Abstract

This paper simulates the multi-layer defense mechanism of biological immune systems to propose a new media public opinion evolution model. Applicable to emergencies and crisis prevention, the model includes five distinct stages: acquisition, analysis, response, disposal, and recovery. By integrating public opinion monitoring and analysis technologies with adaptive evolutionary capabilities, the researchers designed a prototype system that utilizes multi-layer linkage to provide early warning, decision-making support, and situational recovery.

Introduction

In the era of "new media," platforms like Weibo and WeChat allow public opinion to shift and spread with the speed and unpredictability of a biological virus. Traditional monitoring methods are often reactive, focusing on post-event assessment rather than proactive defense. This research identifies three major gaps in current crisis management: a lack of proactive "defensive" warnings, an over-reliance on qualitative descriptions rather than quantitative rigor, and the inability of traditional models to handle the complex "overlay effects" between online and offline social interactions.

To address these challenges, this paper introduces an Artificial Immune System (AIS) model for public opinion. By treating viral misinformation and sudden shifts in sentiment as biological pathogens, the framework develops an adaptive defense to neutralize public opinion "infections" before they destabilize social order.

The Multi-Layered Defense Model

The proposed AIS model simulates the multi-layer defense mechanism of a living organism, structured into five functional layers:

  1. Acquisition Layer: Acts as the "skin" of the system, using search engines and data cleaning to scan for "antigens" (sensitive words and raw data).
  2. Analysis Layer: Extracts "antigenic genes" (malicious hot words) and performs dynamic matching against an "antibody library" to identify threats.
  3. Response Layer: Triggers automatic or manual responses, including public opinion clustering and guidance to counter negative information.
  4. Disposal Layer: The "immune defense" stage. It provides decision support for managers to downgrade threats (e.g., deleting or blocking malicious content) while protecting the integrity of positive information.
  5. Recovery Layer: Handles system restoration and data backup during the "remission period" of a crisis.

Technical Implementation

The prototype system utilizes a sophisticated technical stack to ensure high-speed processing and accuracy:

  • Word Segmentation: Employs the ICTCLAS (Institute of Computing Technology, Chinese Lexical Analysis System) for precise text processing.
  • Clustering: Uses K-means text clustering to group related public opinion threads.
  • Prediction: Utilizes a BP Neural Network optimized by Genetic Algorithms and Simulated Annealing to improve the accuracy of trend forecasting and threat detection.

Empirical Methodology and Results

The study uses Sina Weibo as its primary empirical environment, collecting real-time data via the platform's API. The system tracks forwarding volumes, unique identifiers, and high-activity threads to identify "hot spots."

Key Insights

  • The Viral Metaphor: By applying biological immunology theory to digital governance, the system treats public opinion dissemination as a viral infection, allowing for more dynamic response strategies.
  • Adaptive Evolution: Unlike static keyword filters, the system's "antibody library" evolves as new sensitive topics emerge, allowing it to recognize future mutations of a crisis.
  • Automated Escalation: The model handles minor "infections" automatically in the first three layers. Human intervention is only triggered when a threat reaches the Disposal layer, ensuring responders are not overwhelmed by noise.

Performance Outcomes

The optimized BP neural network showed high precision in identifying Weibo hot topics compared to standard models. Furthermore, the system successfully automated the generation of trend analysis reports, providing the quantitative statistical data necessary for scientific government decision-making. By integrating response and disposal technologies, the prototype demonstrated the ability to significantly reduce the impact of public opinion crises through multi-layer linkage.

Conclusion and Future Work

While the prototype is highly effective for the open API and short-text format of Weibo, its effectiveness on closed-loop platforms like WeChat remains a subject for future study. Additionally, while the system provides high-accuracy discovery, the final disposal of major crises still relies on human-in-the-loop intervention. Future research will explore more autonomous "digital antibodies" capable of guiding public discourse with minimal manual oversight, further increasing the speed of crisis mitigation.

Summary updated

Abstract

This paper simulates the multi-layer defense mechanism of biological immune system and proposes a new media public opinion evolution model applicable to the evolution of public opinions of emergencies and crisis prevention and control, including acquisition layer, analysis layer, response layer, disposal layer and recovery layer. By integrating a variety of public opinion monitoring and analysis technologies, equipment and software development, and then designs a public opinion crisis prevention and control prototype system with adaptive evolutionary capabilities, which is multi-layer linkage combined with public opinion monitoring, public opinion detection, public opinion early warning, public opinion decision-making and public opinion situation recovery.
BibTeX
@misc{yang2017research,
  author = {Kongyu Yang and Ruijie Miao},
  title = {{Research on Evolution Model and Crisis Prevention and Control of New Media Public Opinion}},
  year = {2017},
  doi = {10.2991/iceiss-17.2017.60},
}
Figure