Analysis Q1 2024 toxic bot comments
The dataset of toxic bot comments detected during the first quarter of 2024 included over 4,500 English messages (n=4,597) posted between January 1 and March 31, 2024. The average toxicity score was relatively low at 0.20. However, a notable spike occurred on March 27, 2024, potentially linked to the bombing of Kharkiv, Ukraine, by Russian forces on that day (see the full list of events here).
Despite the low overall toxicity, some messages contained highly problematic content featuring toxic keywords such as nazi, idiot, zionist scum, kill them, jew infestation, degenerate, moron, new world order, scum, and Ioxist—indicating a strong antisemitic narrative.
Approximately 10 messages were classified as very toxic, with scores exceeding 0.8 (on a scale from 0 to 1, where 1 indicates maximum toxicity). About 5% of the messages involved Threatening content, while 3% contained Disinformation. The most common combination of categories was Politics + Religion (see the explanation of the categories here).
Below is a word cloud showcasing the top 50 most frequently used keywords in the dataset from the first quarter of 2024. The size of each word reflects its relative frequency, with prominent terms including black, death, trump, and genocide being among the most frequently used.
These words form the top 50 used keywords by bot comments during Q1 2024
The table below highlights the various toxic categories detected during the first quarter of 2024, along with their respective frequencies (represented by black dots and explained further below). The most prominent categories were Racism, Politics, Religion, and Threat. However, content associated with the Jews category had the highest toxicity score, averaging 0.40 (on a scale from 0 to 1, where 1 indicates maximum toxicity).
The bar chart below shows that the most prominent categories within the toxic bot content during the first quarter of 2024 were Profanity, Ridicule, and Politics. The Contempt category contained the highest concentration of content with a toxicity score of 0.80. The majority of the content, however, fell under the Politics category, indicating that a significant portion of the bot comments aimed to influence political discussions (see the explanation of the different categories here).
Number of toxic messages by category
The keyword network below illustrates the primary focus of toxic bot content during the first quarter of 2024. Different colors represent clusters of words frequently used together, revealing key narratives and associations. One cluster includes terms like immigrant, he, Mexican, and Biden, indicating a focus on topics related to immigration and the U.S. administration. Another prominent cluster features words such as Muslims, Jews, Jewish, whites, and crimes, highlighting discussions centered around racial, religious, and antisemitic themes.
Co-occurring keywords in hate speech in Q1 2024 (racism + threat + contempt)