Stakeholder Context-aware Sentiment, Conversation and Network Analyses in Crisis Communication

This study analyses stakeholder engagement and sentiment patterns during the COVID-19 pandemic across Russell Group universities through Twitter conversations. The research examined over 30,000 tweets across 17,507 conversations from March 2020 to February 2021, using context-aware sentiment analysis, network analysis, and statistical techniques to understand crisis communication dynamics in higher education. The study introduced a novel approach combining logical alignment, sentiment polarity, and semantic similarity analysis to provide a more comprehensive understanding of stakeholder engagement during crisis response. This method revealed complex relationships between emotional tone, topical focus, and agreement levels with university communications. The findings revealed that traditional sentiment analysis alone was insufficient for capturing authentic stakeholder responses, as many replies exhibited neutral stances despite negative tones, suggesting unresolved concerns or general dissatisfaction. Even positive replies did not necessarily indicate genuine agreement with university messages. Through the development of a mathematical ranking system, the research offers a comparative framework for evaluating crisis performance across institutions. The findings emphasize the importance of looking beyond surface-level sentiment analysis and highlight potential challenges in universities' crisis communication strategies. These insights contribute to both theoretical understanding and practical applications for improving stakeholder engagement.
Keywords: Crisis Communication, Stakeholder Behaviour, Neutral Sentiment, Context-aware Sentiment Analysis, Higher Education, NLP, Network analysis, Conversation Analysis.