Data Science/Computational Social Science seminar: Xuan Lu
Understanding Online User Groups Through Emojis
Abstract:
Emojis have been widely used by Internet users to express emotions and enrich user experience. Being adopted in Unicode and supported by many applications, these simple visual ideograms are easy to use and understand, making them popular and “ubiquitous” across different locations, different languages, different platforms, and different cultural backgrounds. Emojis often encode richer emotion and semantics than words, making them efficient carriers of non-verbal cues and contextual information, and they are usually used as complements or surrogates of natural language. With these desirable characteristics, emojis can serve as a novel instrument to study the commonalities and differences of different user groups, when sufficient emoji usage data are available. This instrument is especially beneficial for research problems previously restrained by language or platform barriers. We show that the preference for emojis across user groups can be explained by gender and cultural differences. We show that emojis are able to catch both general and group-specific information by developing emoji representation models in a cross-lingual sentiment classification task. Furthermore, through investigating emoji usage in a specific user group (i.e., developers), we show that emoji usage is correlated with developers’ working status and can be used to track the working status of individual developers and indicate attrition risks.
Speaker Bio:
Xuan Lu is a research fellow in the School of Information, University of Michigan. She earned her Ph.D. in Computer Science from Peking University. She is interested in large-scale user behavior data analysis. Her current research focuses on using emoji as a lens for understanding the languages, sentiments, health, behaviors, and cultural differences of social media users. She is a recipient of the WWW best paper award in 2019 and Microsoft Research Asia Fellowship in 2017.