Layout for automated creation of users (bots) in social networks and imitation of their activity

The article describes methods for automated creation of bot users in the social network VKontakte, imitation of their activity and publication of information in groups, based on the proposed context. The developed algorithm for creating users, used in the methodology, allows you to avoid blocking automatically created users. Methods for simulating activity and publishing information are based on statistical information about the activity of social network users and methods of the VKontakte API.

Using the proposed methods, within the framework of the experiment, bot users were created that publish various information in the groups of the social network VKontakte. According to the results of the experiment, about 17% of the created bot users were blocked, which makes it possible to use the developed layout for information confrontation.

Introduction

At the beginning of 2021, the population of Russia amounted to 145.9 million people, among which 67.8%, or 99 million, are users of social networks, including the social network VKontakte. Compared to last year, the number of users increased by 4.8 million people, which corresponds to 5.1%.

Statistical indicators indicate that in the conditions of modern development, the coverage of social networks is growing faster than the legislation has time to develop special regulations for them. At the moment, there are no effective tools to control compliance with the requirements of legislative acts in the field of the Internet, in particular, in social networks. This leads to the use of social networks and other Internet resources for the dissemination of extremist materials, slander and propaganda of prohibited materials, groups, etc. on the territory of the Russian Federation, which, subject to the anonymization of users on the Internet, makes it very difficult to bring citizens to liability, provided for by regulatory enactments, for the above actions.

Review of existing solutions for automated creation of users and imitation of their activity

Most managers for administering accounts and communities do not have any functionality related to the creation and, moreover, imitation of the activity of bot users. They help manage business accounts and promote communities. But among all the products from Perfect.Studio stand out: VKAccount Manager, VKTalker, VKParser.

If VKTalker and VKParser are not of particular interest regarding the layout being developed, then VKAccount Manager almost completely covers the required functionality:

• simultaneous operation of different accounts;

• registration, unfreezing, activation, password change, cleaning and filling accounts;

• creating, filling out, sending invitations to add to the community / friends, clearing existing invitations from communities / users, clearing posts on the community wall;

• adding to the “Like” list, sending private messages, reposting, adding to friends, posting on the wall and writing comments;

• captcha input support: ruCaptcha, Antigate, manual mode, CapMonster, xEvil;

• proxy support;

• built-in browser;

• text randomization;

• a separate schedule for each task;

• detailed statistics of the program, task and individual account;

• built-in bot for making money on VTope, Likest and other services.

The disadvantages of this software include:

• cost $20 per month;

• the need to use a third-party service for renting phone numbers;

• there is no built-in proxy;

• closed program code;

• lack of imitation of activity according to the given context.

Collection of statistical data on user activity in the social network VKontakte

Paramount in the developed layout is the method of creating bot users, which should provide high resistance to blocking by the VKontakte administration. In order to solve the problem, data were collected on the means of identifying and detecting bots in the social network.

We introduce the concept social bot is a specialized software that is designed to simulate the behavior of a live user in social networks.

The classification of such bots is shown in Figure 2.

You can define the characteristics inherent in bot users:

• account blocking: a live person is highly likely to restore an account when it is easier for the owner of a bot network to create a new account after blocking;

• the link to the page has not been changed: not a very significant sign, but, as practice shows, it is rare to see a real user with an unnamed link;

• lack of an avatar;

• absence from the network for more than 1-3 months: in an age when everyone has a social network client installed on their phone, such low activity seems suspicious;

• subscription to more than 500 communities: a living person is unlikely to have so many subscriptions, since the target news will simply be lost in the “garbage” news feed;

• subscription to many communities from different cities;

• number of friends: both high activity when there are no friends, and thousands of friends, when a typical user can rarely have more than 300 subscriptions, are suspicious. Therefore, it is not the indicator of the number of friends that is interesting, but its relationship with the duration of the account, the number of publications and comments;

• fullness of the profile: as a rule, when creating a bot, either only the required fields are filled in, or absolutely everything, which is extreme, not typical for the average user;

• activity: high activity in filling the profile with content in a short time is a clear sign of a bot.

Also, during the development of the layout and the creation of user-bots, signs of classifying the account as suspicious activity, leading to the blocking of the created account, by the VKontakte administration were revealed:

• using a temporary number when registering an account;

• using a virtual number during registration;

• reason to believe that the profile was created not to communicate with friends and relatives, but for other purposes that are contrary to the rules of the platform.

For the greatest similarity of bot users with ordinary users, it is necessary to resort to statistical data on the activity of users of a particular social network, in particular VK.

To obtain an objective picture of the fullness of the accounts of an average user, statistics were collected: 137437 user IDs were randomly taken, among which the percentage of open, closed and blocked or deleted accounts was identified.

Further, among the open accounts, statistics were collected on the fullness of one or another profile field.

Table 1 – Statistical information about the completeness of the profile

	Minimum	Average	Maximum
Albums	0	2.93	996
Audio recordings	0	135.29	10010
Friends	0	347.17	10000
Number of objects in the block “Interesting pages”	0	56.16	4576
Subscriptions	0	10.22	4472
Video recordings	0	70.87	13089
Posts on the user’s wall	0	169.39	60545

Experimental results

Registration and filling of accounts

Several bot users were created to test the functionality of the developed layout and the ability to avoid blocking an account due to suspicious activity.

Profile blocking prevention statistics is 83%.

Accounts were filled based on the average statistical values ​​obtained in Table 1.

Table 2 – Comparison of the fullness of the profile of an average user and a bot user

	Average user	Bot user
Albums	2.93	3
Audio recordings	135.29	0
Friends	347.17	1
Number of objects in the block “Interesting pages”	56.16	56
Subscriptions	10.22	thirty
Video recordings	70.87	84
Posts on the user’s wall	169.39	16

According to the table above, three key differences can be seen: audio posts, friends, and posts on the user’s wall. There are no audio records in the bot profile due to VK API access restrictions. The number of friends is very different downwards, since the developed layout only sends requests to add to the list of friends to users, but does not affect their decision to accept, reject or ignore this request.

Posts on a user’s wall have a publication date, and adding a huge number of posts at once increases the likelihood of a bot user being exposed by regular users. In connection with the above, the number of entries on the user’s wall will gradually increase, just like a regular user.

Publication of information in social network groups

In order to try to influence the opinions of people in VKontakte communities, two experiments were carried out:

• publishing comments under posts that relate to a specific topic in a popular multimillion-dollar group;

• posting neutral, off-topic comments under various posts in a group of about 100 people.

The first experiment yielded results: several “I like” marks and positive responses. At the same time, comments were quickly lost in a large stream of new ones.

The second experiment did not cause any reaction, although the number of bots used was approximately 10% of the entire community.

conclusions

When developing a software tool for simulating user activity, an undocumented VK API method was found and used. The conducted experiments in the field of filling accounts showed the correctness of the proposed methods and the success of creating bot users (83% of accounts remained active after the experiments) and simulating their activity. Problematic issues were found when working with audio recordings, which are currently not available in the VK API. At the same time, it should be borne in mind that it is not the statistics of fullness that is important, but its relationship with the duration of the account.

Experiments related to the publication of information in communities give mixed results. On the one hand, in popular groups, when contextual search is connected, there is a visible result. But on the other hand, in such conditions, a huge network of bots is required so that you can count not only on a few “I like” marks, but also on the impact on the opinions of others. At the same time, attempts to attract attention in a small group did not give any result. The true reasons are unclear, but several hypotheses can be put forward:

• small communities are reluctant to accept unknown members;

• neutral, but different from the theme of the group, the content does not attract the attention of such a team;

• insufficient attention of community members to comments under posts.

Promising areas for development include:

• adding the ability to block / remove unwanted community / content and freeze the user’s page;

• expansion of the network of bot users, as this will successfully affect the possibilities of changing the opinion of the community, propagating ideas and other similar goals;

• the use of artificial intelligence technologies (for example, ChatGPT4) to automate the formation of bot user comments on a specified topic.