Jointly Learning to Detect Emotions and Predict Facebook Reactions

Graziani, Lisa; Melacci, Stefano; Gori, Marco

doi:10.48550/arxiv.1909.10779

Cited by 1 publication

(1 citation statement)

References 0 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…A similar approach was made by Raad et al (2018) to create a framework for predicting the reaction distribution on Facebook posts. They examined the potential Facebook reaction usage, recognized classified emotions, and found the result through their dataset, which revealed that “Like,” “haha,” and “love,” were the most frequently used “Facebook reactions.” Graziani et al (2019) proposed the task of emotion classification and prediction of “Facebook reactions” and warned that Facebook reactions could easily become ambiguous in mapping emotion classes. Later, Wisniewski et al (2020) identified many design constraints of the “Facebook reaction” feature and users’ inability to express conflicting emotions.…”

Section: Discussionmentioning

confidence: 99%

Facebook reactions in the context of politics and social issues: a systematic literature review

Anwar,

Giglietto

2024

Front. Sociol.

View full text Add to dashboard Cite

In February 2016, Facebook expanded the original “Like” button by introducing five additional “Reactions”—Love, Haha, Wow, Sad, and Angry—using modified versions of Unicode emojis. These reactions enable users to express more nuanced emotions towards posts. This literature review investigates scholarly research on user behavior in response to these reactions, with a focus on a broad spectrum of socioeconomic and psychological issues. We conducted a systematic search across databases including Scopus and Google Scholar, using keywords such as “Facebook” and “Reaction,” combined with various key phrases and Boolean operators. Our review synthesizes sixty-four articles published from 2016 to 2023, exploring diverse topics such as political news, far-right and extremist parties, racism, and hate speech during the COVID-19 pandemic. We organized these articles by theme and publication date. Our meta-analysis reveals that lifestyle and entertainment posts predominantly receive positive reactions, while sociopolitical content tends to elicit a broader spectrum of emotions, including negative sentiments. Furthermore, emotionally charged content consistently attracts higher volumes of reactions, regardless of sentiment. This research highlights the intricate relationship between user reactions and content characteristics, providing deeper insights into the dynamics of online engagement. By understanding these interaction patterns, we gain a better grasp of emotional responses and engagement levels, which ultimately shape online discourse and user interactions.

show abstract

Section: Discussionmentioning

confidence: 99%