Huitema, Imme (2026) Analyzing Echo Chambers on the Decentralized Web: a Mastodon Case Study. Master's Thesis / Essay, Artificial Intelligence.
|
Text
MasterThesisImmeRientsHuitemaUniversityofGroningen.pdf Download (21MB) | Preview |
Abstract
Echo chambers on social media are often studied on centralized platforms, where governance and recommendation algorithms shape exposure to users and content. Less is known about how these dynamics emerge on decentralized platforms such as Mastodon, which consists of independently governed instances and lacks engagement-driven recommendation algorithms. This thesis examines echo chamber dynamics on Mastodon through network connectivity, sentiment alignment, and content spread. We analyze selected discussions on mastodon.social using structural network measures, follower overlap, and sentiment analysis. We also investigate smaller topic-based instances by examining follower origins, boosting behavior, and the instance diversity of communities in follower and boost networks. The results show that echo chamber dynamics are not uniform. On mastodon.social, low clustering and limited follower overlap provide little evidence for strong structural echo chambers, while discussion threads show strong sentiment alignment. In the smaller-instance datasets, follower relations are locally concentrated, but boosts frequently cross instance boundaries and communities remain diverse in their instance composition. Overall, Mastodon shows some echo chamber-like tendencies, but these vary across network layers. Local concentration shapes follower relations, while cross-instance boosting supports content diffusion beyond instance boundaries.
| Item Type: | Thesis (Master's Thesis / Essay) |
|---|---|
| Supervisor name: | Riveni, M. and Grossi, D. |
| Degree programme: | Artificial Intelligence |
| Thesis type: | Master's Thesis / Essay |
| Language: | English |
| Date Deposited: | 24 Jun 2026 13:22 |
| Last Modified: | 24 Jun 2026 13:22 |
| URI: | https://fse.studenttheses.ub.rug.nl/id/eprint/37326 |
Actions (login required)
 |
View Item |