Scholarlink

<p>During public health crises and disaster events, comprehending the evolution of public concern is critical for effective risk communication and policy-making. Given this imperative, harnessing social media data offers a real-time lens into a community’s collective response. However, traditional analytical approaches fail to capture the complex, dynamic nature of the discourse. To address this challenge, we define <i>community concern</i> in context to an event as a multi-dimensional construct, characterized by the event’s <i>attentional salience</i> and its <i>sentiment impulse</i>. To quantify these dimensions, we introduce the <i>CUE-ESI</i> framework, formulating two novel metrics: <i>Concern Uptake and Engagement</i> (CUE) and <i>Effective Sentiment Impulse</i> (ESI). This framework transforms social media data into a structured representation of these metrics, enabling us to discover recurring concern states and predict them. To validate the framework, we apply it to social media data that originated from an urban community during the COVID-19 pandemic. Our unsupervised analysis identified four concern states: <i>Dormant Concern</i>, <i>Turbulent Concern</i>, <i>Emerging Positive Concern</i>, and <i>Escalating Negative Concern</i>. Qualitative validation of these states revealed that they aligned with ground-truth events. The state classification model predicted these states with an <InlineEquation ID="IEq1"> <EquationSource Format="TEX">\(F1_{micro}\)</EquationSource> </InlineEquation> of 0.97. Further, qualitative prediction on future data demonstrated the framework’s real-world relevance and its capability as an early-warning system. This robust and interpretable framework can contribute to enhancing situational awareness and shaping more effective and timely public health communication strategies.</p>

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

From discourse to dynamics: the CUE-ESI framework for modeling community concern

  • 期刊论文

  • 发表日期：

    2026年1月30日

查看全文
  • Divya Gupta,
  • Shampa Chakraverty

摘要

During public health crises and disaster events, comprehending the evolution of public concern is critical for effective risk communication and policy-making. Given this imperative, harnessing social media data offers a real-time lens into a community’s collective response. However, traditional analytical approaches fail to capture the complex, dynamic nature of the discourse. To address this challenge, we define community concern in context to an event as a multi-dimensional construct, characterized by the event’s attentional salience and its sentiment impulse. To quantify these dimensions, we introduce the CUE-ESI framework, formulating two novel metrics: Concern Uptake and Engagement (CUE) and Effective Sentiment Impulse (ESI). This framework transforms social media data into a structured representation of these metrics, enabling us to discover recurring concern states and predict them. To validate the framework, we apply it to social media data that originated from an urban community during the COVID-19 pandemic. Our unsupervised analysis identified four concern states: Dormant Concern, Turbulent Concern, Emerging Positive Concern, and Escalating Negative Concern. Qualitative validation of these states revealed that they aligned with ground-truth events. The state classification model predicted these states with an \(F1_{micro}\) of 0.97. Further, qualitative prediction on future data demonstrated the framework’s real-world relevance and its capability as an early-warning system. This robust and interpretable framework can contribute to enhancing situational awareness and shaping more effective and timely public health communication strategies.