Public Opinion Crisis Management via Social Media Mining

Yu Ma, Rui Mao, Peng Wu, Erik Cambria

Abstract: Enterprises suffer tremendous economic and reputational losses in public opinion crises. However, providing specific guidance for enterprises to respond effectively to public opinion crises is challenging. This study proposes a netizen-centered approach to identify crisis response opportunities for enterprises by mining social media data. First, we identify the topics discussed by netizens with the Biterm Topic Model, thereby evaluating their importance levels. Then, the negativeness levels of crisis topics are quantified by sentiment analysis. Lastly, crisis response opportunities are quantitatively identified and prioritized by applying an opportunity algorithm that simultaneously considers each crisis topic's importance and negativeness levels. Experimental results demonstrate the validity and superiority of our approach. Moreover, we demonstrate that a very negative and less important topic in the growth stage will likely become very negative and important at the maturity of the crisis. This finding supports decision-making in response to critical topics at the early stage of a crisis, preventing the deterioration of the public opinion crisis.

External IDs:dblp:journals/taffco/MaMWC25