Understanding Inorganic Behavior on Twitter: a Machine Learning Approach for Journalistic Applications

Thesis Project created by Alejandro Lozada English
1y ago update
Twitter is a micro-blogging social network in which users post, like, promote and share short written messages, images, links, or videos known as “tweets”. The platform is amongst the most popular social networks in the world, with an estimate of over 400 million users for 2022.

The existence of bots on Twitter has been widely the subject of research. Documenting its presence on the platform, understanding the different categories in which they can be studied, and developing detection techniques and algorithms has been the main focus of several studies.

In recent years, the influence bots have on the attention and information dynamics in media ecosystems with heterogeneous agents has also awakened interest in both academia, journalism and civil organizations. Bridging the gap between bot detection on Twitter and the study of their influence in media ecosystems has become more relevant than ever. As many discussions of public interest start on the platform, enabling journalists and civil organizations to do a better job reporting on issues happening there is crucial for the well-being of liberal democracies where social media is becoming the default public forum.

The aim of this Master’s thesis is to determine the feasibility of the creation of a tool
for journalists and civil organizations to detect and better understand malicious activity on Twitter.

In this context, “malicious activity” is considered to be any attempt to mislead public opinion using inorganic behavior and exploiting Twitter’s social-network features.


Data, Technology
Civil Society, Media, Research
Prediction, Language
Explainability and Interpretability, Machine Learning, NLP


2023-03-01 12:55 Weblink
This is the open repo for the project