Identifying accounts that spread disinformation on Twitter

Adil Kussataiuly

Bachelor's thesis

Abstract:

The spread of disinformation on social media platforms, particularly Twitter, has become a growing concern in recent years. This study aims to identify networks of Twitter accounts that spread disinformation by analysing a dataset of tweets related to specific global actions, such as elections, COVID-19, and war. Over 10,000 users were randomly selected from the dataset, and the Botometer® algorithm was used to identify potential bots. Various methods were used to detect Twitter accounts spreading disinformation, including network analysis, natural language processing, and machine learning algorithms. The Botometer® algorithm was selected as the computational method for bot identification in this study. A unique dataset was created by searching for particular keywords related to each topic of interest. The findings suggest that the number of tweets is a valuable variable in distinguishing low-activity bots from highly active users in the network. The results also showed that the group of users who tweeted about the election had the highest number of tweets but the lowest spammer value, while the groups of users who tweeted about COVID-19 and war had significantly fewer posts. This study confirms the relationship between bot score and the number of fake subscribers/spammers, providing valuable insights for detecting and preventing fraudulent behaviour on various online platforms. The practical implications of this research include improving the accuracy of bot detection systems and developing better tools for detecting and preventing disinformation on social media platforms. Ongoing research efforts in this area provide promising methods for mitigating the spread of disinformation and improving the accuracy and reliability of online data.

Keywords

Twitter, spread, elections, propaganda, fake, COVID-19, accounts, bots

Language used: English

Date on which the thesis was submitted / produced: 24. 3. 2023

Thesis defence

• Date of defence: 16. 5. 2023
• Supervisor: Ing. Miloš Ulman, Ph.D.
• Reader: Tomáš Vokoun, Ing.

Czech University of Life Sciences Prague

Faculty of Economics and Management

Bachelor programme:
Informatics