Information can be extracted even from sets of stickers
A neural network has been developed in Russia, which is designed to identify administrators and owners of anonymous channels by studying them in Telegram. This was reported by the publication «Izvestia» with reference to the press service of the National Technology Initiative (NTI) and the developer company T.Hunter.
It is noted that the neural network called «Comrade Major» acts similarly to «human» analytics, checks the channel description, information in messages and built-in chats, but does it faster and with better quality. It is expected that it will be used by government or private organizations involved in the investigation of cybercrimes.
Igor Bederov, head of the information and analytical research department at T.Hunter, said:
A basic anthology of methods and techniques that a specialist uses when researching a Telegram channel was developed, and then the program was taught these techniques.
He added:
AI is based on one of the most common text neural networks, which runs on our company’s servers. A Telegram bot has been written to use it, which simplifies interaction with the software.
The neural network collects and studies a lot of data, including information about administrators from the channel description, its archived copies, messages, metadata of published files, message signatures, built-in chat, sticker packs, documents and videos, and so on. & nbsp;
Also, the neural network can check legal entities by TIN and provide general information, including period of existence, number of employees, authorized capital, registration address. The developers plan to teach the software to explore social media accounts, phone numbers, email addresses and cryptocurrency wallets.