Abstract

The increasing prevalence of location sharing features on social media has enabled researchers to ground computational social science research using geolocated data, affording opportunities to study human mobility, the impact of real-world events, and more. This paper analyzes what crucially separates tweets with geotags from tweets without. Our findings show that geotagged tweets are not representative of Twitter data at large, limiting the generalizability of research that uses only geolocated data. We collected 1.3M geotagged tweets on Twitter (most of which came from Instagram), and compared them with a random dataset of tweets on three aspects: affect, content, and audience engagement. We show that geotagged tweets on Twitter exhibit significantly more positivity, often citing joyous and special events such as weddings, graduations, and vacations. They also convey more collectivism by using more …

Metadata

publication

WebSci '23: Proceedings of the 15th ACM Web Science Conference 2023, 136-146, 2023

year

2023

publication date

2023/4

authors

Julie Jiang, Jesse Thomason, Francesco Barbieri, Emilio Ferrara

link

https://dl.acm.org/doi/abs/10.1145/3578503.3583596

resource_link

https://dl.acm.org/doi/pdf/10.1145/3578503.3583596

conference

WebSci '23: Proceedings of the 15th ACM Web Science Conference 2023

pages

136-146

publisher

ACM