Jun 03, 2021 16:00:00

It turns out that nearly half of Twitter's trends were artificially manipulated by bots

Twitter has a 'trend ' feature that shows you what's hot in a particular region or country. A research team at the Swiss Federal Institute of Technology Lausanne (EPFL) investigated the relationship between trend functions and bots, and found that 47% of Turkish trends and 20% of global trends were artificially created by bots. Turned out.

Ephemeral Astroturfing Attacks: The Case of Fake Twitter Trends
(PDF link) https://arxiv.org/pdf/1910.07783.pdf

Mass scale manipulation of Twitter Trends discovered --EPFL

https://actu.epfl.ch/news/mass-scale-manipulation-of-twitter-trends-discov-2/

It is known that there are a large number of bots on Twitter that automatically post tweets disguised as humans, and a 2017 study also revealed that as many as 15% of active Twitter users are bots. I will. The research team investigated the impact of these bots on Turkish and global trends.

Research results show that up to 48 million Twitter users are bots-GIGAZINE

The figure below shows the 'percentage of trending things artificially trended by bots' in Turkey (upper graph) and the world (lower graph), with the bar graph ranking 10 Within the rank, the line graph shows the percentage of things that are within the top 5 in the ranking. Looking at the graph, bots averaged 47.5% of the top 5 trends in Turkey, and 19.7% of the top 10 trends in the world were artificially created by bots. You can see that.

The graph below shows the 24-hour distribution of things that are trending by bots (red) and things that are not related to bots (blue). Looking at the graph, we can see that the bot's activity becomes active in the afternoon and peaks at 8 pm. 'Because people often watch Twitter at night, bot attackers may be focusing their attacks at night,' the research team points out.

According to the research team, artificial manipulation of Twitter trends can have a significant impact on society. The hashtag '#SuriyelilerDefolsun', which is known to have actually been trended by bots, has been featured in several news and academic papers.

The research team told Twitter that the trend was being manipulated by the bot, but at the time of writing the article, the problem was not fixed.