Analysis Q2 2024 toxic bot comments
The dataset of toxic bot comments detected in the second quarter of 2024 included over 4,300 English messages (n=4,333), posted between April 1 and June 30, 2024. The average toxicity score was relatively low, at 0.20. However, a significant spike occurred on April 17, 2024, possibly linked to key events that day, such as the Russian invasion of Ukraine, the Iran-Israel conflict, the Israel-Hezbollah conflict, or the war in Sudan (see the full list of events here).
Despite the low average toxicity, several messages featured highly problematic language, with toxic keywords including idiot, moron, filthy Jew, Israhell, bitch, scum, tards, new world order, and demoncrats.
5 messages were classified as very toxic, with scores exceeding 0.8.
Toxicity often involved Ridicule (15%), Politics, Contempt, and/or Religion.
5% of the messages contained Threatening content, while 2% involved Disinformation.
The most frequent category combination was Politics + Religion (see category explanations here).
Below is a visualization of the top 50 toxic keywords used during the second quarter of 2024. The size of each word reflects its frequency relative to others, with terms like genocide being the most common, followed by Israel, black, and killed.
These words form the top 50 keywords used by bot comments during Q2 2024
The table below highlights the various toxic categories detected during the second quarter of 2024, along with their respective frequencies (represented by black dots and further explained below). The most prominent categories were Ridicule, Politics, Religion, and Threat. However, the Jews category had the highest average toxicity score of 0.40 (on a scale from 0 to 1, where 1 represents the highest level of toxicity).
The bar chart below highlights the most prominent categories within the toxic bot content from the second quarter of 2024. Profanity, Ridicule, and Politics were the leading categories, but the Politics category had by far the most content, indicating that a significant portion of the bot comments focused on influencing political discussions.
Number of toxic messages by category
The keyword network below visualizes the primary focus of toxic bot content from the second quarter of 2024. Different colors represent clusters of words frequently used together, revealing key narratives promoted by the bot comments. A significant cluster centers around terms like Israel, die, terrorist, genocide, killed, crimes, and Jewish, highlighting discussions related to violence, conflict, and antisemitic narratives.
Trends
The analysis of toxic bot comments detected in the first and second quarters of 2024 reveals several similarities and differences:
Similarities:
Consistently Low Toxicity Scores: In both Q1 and Q2, the average toxicity scores remained relatively low (0.20).
Focus on Israel and Jews: A major portion of the bot content in both quarters revolved around antisemitic narratives, as shown by the frequent appearance of terms related to Israel, Jews, genocide, and Zionism.
Similar Category Scores: The distribution of toxicity across categories like Politics, Ridicule, and Religion showed consistent patterns across both quarters.
Overlapping Keywords: Many of the most frequently used toxic keywords, such as genocide, terrorist, black, and Israel, appeared prominently in both the word clouds and keyword networks.
Differences:
Increased Political Focus in Q2: The second quarter showed a greater emphasis on political content, with Politics receiving a larger share of toxic bot comments compared to Q1.
Doubling of Threatening Content: The proportion of messages containing Threats doubled, from 5% in Q1 to 10% in Q2, indicating an escalation in harmful and aggressive language.
Shifts in Toxic Narratives:
Q1: Toxic narratives focused more on racism and racial conflicts, as indicated by the prominence of keywords related to race in the word cloud.
Q2: The narrative shifted toward genocide and violent conflict, with a stronger focus on terms related to violence and mass harm.
Additional Narrative in Q1: Apart from the focus on Israel and Jews, Q1 also displayed a clear narrative centered around Biden and immigrants, which was less apparent in Q2.
Overall, the analysis indicates that while there is continuity in the toxic narratives between the two quarters, Q2 saw an escalation in politically charged and threatening content, with a shift in focus from racism to genocide and violence.