The Pulse of Mood Online: Unveiling Emotional Reactions in a Dynamic Social Media Landscape

TL;DR

The paper develops an unsupervised pipeline to study collective emotional and moral reactions to offline events on social media by constructing daily time series of aggregate affect from transformer-based emotion and morality labels, detecting change points with combined CUSUM and BOCPD methods, and explaining changes through BERTopic-derived topics. Applied to three large Twitter datasets (2020 LA, 2022 US abortion debates, and 2022 French election), the approach detects both major and subtle events and reveals complex, topic-specific emotional dynamics, including short-term and long-term effects. Evaluation shows robust detection and coherent topic explanations, highlighting the method's ability to disaggregate broad sentiment into nuanced, event-specific narratives. The work demonstrates practical potential for real-time sensing of population reactions during crises and for informing policy communication, with generalizability to platforms beyond Twitter and to multilingual contexts.

Abstract

The rich and dynamic information environment of social media provides researchers, policy makers, and entrepreneurs with opportunities to learn about social phenomena in a timely manner. However, using these data to understand social behavior is difficult due to heterogeneity of topics and events discussed in the highly dynamic online information environment. To address these challenges, we present a method for systematically detecting and measuring emotional reactions to offline events using change point detection on the time series of collective affect, and further explaining these reactions using a transformer-based topic model. We demonstrate the utility of the method by successfully detecting major and smaller events on three different datasets, including (1) a Los Angeles Tweet dataset between Jan. and Aug. 2020, in which we revealed the complex psychological impact of the BlackLivesMatter movement and the COVID-19 pandemic, (2) a dataset related to abortion rights discussions in USA, in which we uncovered the strong emotional reactions to the overturn of Roe v. Wade and state abortion bans, and (3) a dataset about the 2022 French presidential election, in which we discovered the emotional and moral shift from positive before voting to fear and criticism after voting. The capability of our method allows for better sensing and monitoring of population's reactions during crises using online data.

Figures (9)

• Figure 1: Pipeline to detect and measure online emotional reactions.
• Figure 2: Time series of emotions and moral sentiments in Los Angeles Tweets from January 1 to August 1, 2020. We show the daily fraction of tweets with different affect labels. The notable peaks and dips in the time series can be associated with the external events marked as vertical lines.
• Figure 3: Top events, their associated topics and corresponding emotional and moral reactions detected in 2020 Los Angeles data. See the full list of events in Table \ref{['tab:la_reactions']} in Appendices.
• Figure 4: Short-term and long-term changes of emotions and moral sentiments around four events in 2020 Los Angeles data. Asterisks indicate significance values: * (p-value < 0.05), ** (p-value < 0.01), *** (p-value < 0.001), and no asterisk indicates p-value $\geq$ 0.05.
• Figure 5: Emotions and moral sentiments expressed in COVID-related topics during the two weeks after WHO announcement of the pandemic on 2020-03-11. The topics are COVID ("coronavirus, corona, virus"); grocery panics ("grocery, groceries, shelves", "water, dasani, hydro", and "toilet, paper, rolls"); leisure activities ("episode, episodes, show", "cook, cooking, cookout","tickets, ticket, selling"); and education ("teachers, students, learning", "schools, lausd, classes", "schools, lausd, closed").