Volcanology in the Twittersphere
Jamie I. Farquharson
Affiliations: Institute for Research Administration, Niigata University, Ikarashi 2‑8050, Nishi‑ku, Niigata, 950‑2181, Japan
Presentation type: Poster
Presentation time: Monday 16:30 - 18:30, Room Poster Hall
Poster Board Number: 140
Programme No: 3.1.32
Abstract
The social media service X, formerly and popularly known as Twitter, is a microblogging and social networking service whose users can post, interact with, and re-post messages ("tweets"). Each tweet (object) is associated with an array of data, including a unique identifier, a timestamp, text content, attached images, URLs, and location information. This contribution comprises a multilingual longitudinal study of volcano-centric tweet strings over three years. Tweets containing predefined strings (e.g. "volcanic eruption," "erupción volcánica," or "火山噴火") were crawled and downloaded daily over August 2019--June 2023 using a custom Python script, yielding a dataset of over 12 million tweet objects across 20 languages. To compare datasets with varying data volumes (some languages are much more highly represented on the platform than others), I define a "Tweet anomaly score" α. Spikes in α generally reflect real-world volcanic events, so analysis of the evolution of α over time yields some key insights. For example, the recent Hunga Tonga--Hunga Haʻapai eruption is immediately identifiable in all studied strings: discourse about the eruption transcends national and linguistic boundaries. This and other large or newsworthy eruptions (especially if clustered in time) can bring about a step-change in the amount of online discourse about volcanoes in general. However, analysis also reveals linguistic imbalance: English dominates, even regarding eruptions in countries where English is not the main language. As a result, not all eruptions result in equal online traffic, even if the physical characteristics of the eruptions are themselves similar.