Counteracting disinformation
9 min

Unreal me: fake accounts on the pages of Ukrainian media

showcase image
Counteracting disinformation

About the project

Note: This project was conducted as part of Artellence and, together with similar projects, became a prerequisite for the creation of the NGO Association for the Development of Big Data and Information Technology of Ukraine with the involvement of experts who directly implemented this project.

Active Facebook users state that they use this social network as a source of information because the latter is aggregated there. There is no need to visit various websites if all news media have their pages in Facebook.

The purpose of the study is to investigate how bots shape public opinion in the Ukrainian media.

Why this is important?

A number of factors influence which posts a user sees first in their feed, and one of them is the number of comments. What if the comments are not from real people and the interest in the topic is artificially stirred up? Then instead of news that is interesting to many people, the user is more likely to see news that is promoted by certain algorithms, and their opinions will be formed on the basis of inaccurate information.

Research methodology

Research period: from 01.05.2019 to 08.07.2019.

  • we analyzed the most commented by bots Facebook posts on the pages of the most popular media. The media sample was formed on the basis of the list of the most popular Ukrainian media, to which the pages with the largest number of followers from Ukraine were added. In total, the sample included 332 Facebook pages;
  • we analyzed the top five most popular, regularly read and trusted media outlets:
    • Segodnya (39.8% of comments are from bots),
    • Obozrevatel (34.7%), UNIAN (30.6%),
    • Ukrayinska Pravda (28.8%),
    • Politeka (22.7%);
  • we analyzed comments on the pages where bots were most active, which mentioned Volodymyr Zelenskyi, Petro Poroshenko, Yulia Tymoshenko, Volodymyr Groysman and Svyatoslav Vakarchuk. We categorized these comments by connotation (positive - neutral - negative);
  • we grouped bots’ comments under the most resonant news in the message. A message is the main idea that a certain number of bot comments contains. Each message includes posts that were written in different wording but contained the same meaning. We described only those messages that were included in more than 50 bot comments.

How bots were identified:

  • o identify bots, Artellence developed a machine learning algorithm that analyzed public information from the Ukrainian segment of Facebook for a period of 9 months, from November 2018 to July 2019;
  • both comments and information from profiles were taken into account;
  • only users who left more than 10 comments on political topics were included in the analysis;
  • as a result, we consider those users whom the model classified as bots with a 95% probability to be bots (we used 8 characteristics of bots, which are described in detail in the Project "Erase it if you can. How Ukrainian bots live on the pages of Ukrainian politicians").

How the news was selected:

  • the news was selected using NLP Word Vectors, a natural language processing method of machine learning. The algorithm analyzed the texts of the posts and grouped similar news items into one news item.

Results of the study

Activity of bots on Facebook pages:

  • bots were the most active on the pages of little-known media outlets with a small number of subscribers (up to 200 thousand). More than half of all comments on these pages were made by bots;
  • in the top three:
    • (56% of bot comments, 24,832 followers),
    • Ukraine24 (53% of all comments by bots and 169,659 followers),
    • Ukrainian news (52% of comments are posted by bots, 6,970 followers);
  • у топ-20 сторінок, де кількість коментарів ботів перевищує 45%, увійшли 2 сторінки популярних українських медіа: 
    • RBC-Ukraine* (399,256 followers),
    • Strana.UA (78,984 followers);
  • the top 20 in terms of the number of fake comments also included 15 foreign pages with a large Ukrainian audience (the number of bot comments on them did not exceed 31%). On average, bots’ comments on such resources make up 18%):
    • 5 international news agencies,
    • 9 media from russia (including 6 large well-known media),
    • one media outlet from Israel;
  • the highest ratio of bot comments to all comments on pages:
    • CGTN in Russian (31%),
    • Chastime (29%),
    • DW's Ukrainian (27%) and Russian (25%) divisions;
  • he lowest percentage of bots in comments:
    • Esquire Russia (8%),
    • Nozh magazine (5%);
  • on a third of the media outlets in our sample (115 out of 332 pages), more than 30% of comments are made by bots. This includes 10 media pages from the top 20 media outlets with the largest reach:
    • RBC-Ukraine** (44% of the comments on the page are from bots),
    • Strana.UA (43%),
    • Channel 112 (41%),
    • Segodnya (40%),
    • Gordon (40%),
    • (38%),
    • Censor.NET (35%),
    • Obozrevatel (35%),
    • ICTV Facts (34%),
    • UNIAN (31%).

Behavior of bots:

  • bots are actively commenting on posts on the pages of media with a large audience, but they are most likely to "boost" views on the pages of lesser-known media;
  • bots can comment on posts to create the illusion of relevance and interest in a particular topic;
  • the purpose of creating bots is ideological; they try to shape a certain agenda and thus influence public opinion.

Bots on the pages of trusted media outlets:

  • the top five most popular, regularly read and trusted media outlets are the newspapers:
    • Segodnya (39.8% of comments are from bots),
    • Obozrevatel (34.7%), UNIAN (30.6%)
    • Ukrayinska Pravda (28.8%);
    • Politeka (22.7%);
  • we analyzed the comments on these pages that mentioned Volodymyr Zelenskyy, Petro Poroshenko, Yulia Tymoshenko, Volodymyr Groysman and Svyatoslav Vakarchuk. We categorized these comments by connotation (positive - neutral - negative):
    • or each of the politicians studied, the majority of bot comments were critical and negative (on average, 18% of all comments for which we identified connotations),
    • bot comments with neutral connotations are on average more than 2.5 times less than hateful ones: 6.8% vs. 18%. The number of positive comments is three times less than the number of negative ones - 5.7% on average,
    • bots wrote almost one and a half times more negative and neutral comments about Volodymyr Zelenskyy than about Petro Poroshenko. Positive comments from bots about Zelenskyy were almost five times as many as about the former president (21% positive about Zelenskyy vs. 4% positive about Poroshenko),
    • Vakarchuk is mentioned on only three websites (Segodnya, Politeka and UNIAN) and mostly negatively — 77% of all bot comments about him were critical,
    • although bots on these resources write relatively little about Yulia Tymoshenko (only 4.4% of bot comments refer to Ms. Tymoshenko), they are mostly positive. It makes up 58% of all bot comments about her on these websites,
    • bots write almost nothing about Volodymyr Groysman on the pages of these media outlets.

Bots in comments under high-profile news:

  • he most discussed news in the media in May-July 2019 were socio-political reports: the LGBTQ+ march and Russia's return to PACE;
  • the share of bots in comments is led by news about public policy: Groysman's statements, Poroshenko's raprotests, Tymoshenko and Groysman's dispute on a TV channel (there is a separate study on this "Erase it if you can. How Ukrainian bots live on the pages of Ukrainian politicians");
  • LGBT march:
    • he news caused the biggest public outcry. It garnered 21,640 comments, including 15,954 from real people and 5,686 (26%) from bots,
    • we identified 635 main messages in bot comments. There are few of them compared to other top news stories, as they are more uniform and concentrated. All of them were negative, with varying degrees of aggression,
    • most of the comments were with the text "I do not support LGBTQ+" (180 comments out of 635), 150 comments were against the column of LGBTQ+ military. The third place among bot comments on this news was taken by a message in the style of "Orientation is everyone's own choice, I'm not against LGBTQ+, but the march is too much." There are 52 such messages;
  • Russia returns to PACE:
    • it is the second most resonant news. Bots also accounted for 26% of comments on the news (2,781 bot comments versus 8,078 comments from real people),
    • for 37.5% (1,079) of bot comments, we identified the main message. They mostly criticized Europe (312 comments out of 1,079),
    • messages such as 'Europe is corrupt' (200 bot comments), 'The PACE delegation must be changed!' (158 messages) and 'We must leave PACE and quarrel with Europe' (112 messages) dominated,
    • this news was followed by many messages criticizing the actions of the current (217 comments) and former (90 comments) presidents.

Additional materials


1. Publish the results of this study;
2. Conduct such surveys at regular intervals to check the real public opinion and to allow for the formation of public opinion based on real data;
3. Publishing the results can have a positive impact on reducing disinformation online, as Ukrainian society will have a place to see which comments are fake and which are real;
4. Conduct such studies for other categories of pages


Other projects




Counteracting disinformation
See more