{"id":1570,"date":"2024-10-24T14:27:53","date_gmt":"2024-10-24T11:27:53","guid":{"rendered":"https:\/\/factcheck.lt\/?p=1570"},"modified":"2024-10-24T21:52:21","modified_gmt":"2024-10-24T18:52:21","slug":"","status":"publish","type":"post","link":"https:\/\/factcheck.lt\/eng\/news\/youtube_elections_july2024\/","title":{"rendered":"Fact-checking and debunking disinformation in Belarusian YouTube videos about elections: July 2024","raw":"Fact-checking and debunking disinformation in Belarusian YouTube videos about elections: July 2024"},"content":{"rendered":"

As our previous publications<\/a> and monitoring efforts have shown, the topic of elections around the world remains consistently relevant in the Belarusian information space. In many cases, media outlets, primarily state-owned, deliberately spread false information (disinformation) about elections, especially in democratic countries. Therefore, debunking election-related misinformation is a constant task. We conduct fact-checking and verification using AI.<\/p>\n

\n

Methodology<\/strong> We used a dataset<\/a> prepared by bloggers about Belarusian YouTube channels in the third quarter of 2024. Technically, it contains information about 300+ channels, on which over 24,000 videos were uploaded, garnering more than 800 million views.<\/p>\n

The time period chosen was July 2024<\/strong>.<\/p>\n

Then, a set of videos containing election-related information in their descriptions was filtered. The selection included 65 videos that gathered about two million views.<\/p>\n

Using large language models (LLMs), the text was transcribed and then analyzed for disinformation narratives via the model API – gpt-4o-mini-2024-07-18.<\/p>\n

23.2 hours of video were transcribed.<\/p>\n

The ten channels that gathered the most views are:<\/p>\n

    \n
  1. NEWS.BY: Belarus and World News 910,313<\/li>\n
  2. CTVBY 409,949<\/li>\n
  3. Real Belarus 220,532<\/li>\n
  4. ONT Channel 72,621<\/li>\n
  5. No Guarantees 59,387<\/li>\n
  6. SB TV 59,250<\/li>\n
  7. Belta News Agency 37,853<\/li>\n
  8. USUAL MORNING 34,958<\/li>\n
  9. Pavel Latushka 32,620<\/li>\n
  10. DW Belarus 29,553<\/li>\n<\/ol>\n

    Word cloud after removing filler words and interjections: \"\"