The center for innovative technologies "Boriviter" has initiated trials of the "ODCR Assistant" neural network, designed to simplify the process of passive information collection and analysis from enemy open sources. According to "Militarniy", this project is implemented by the center's analytical department and targets the automatic detection of useful information from the adversary's media landscape.
The module operates with unclassified data streams, including Russian military social media and technical blogs. The system processes approximately a thousand messages daily. Algorithms categorize these messages using "green" and "red" markers. If a message contains "green" keywords and lacks "red" ones, it advances for observation formation.
The next phase will involve verifying the novelty of the data and their systematization. Meanwhile, developers note that complete automation of the analysis remains unattainable, and human involvement is still necessary.
In addition to the "ODCR Assistant", the team is also working on a language model for Ukrainian military personnel, which will operate based on neural networks and feature an interface similar to ChatGPT. The aim is to facilitate dialogue with experts and simplify situational analysis.
The development is carried out by the "Boriviter" team on powerful servers. The organization is open to collaboration but emphasizes the importance of non-commercial support focused solely on the needs of the Ukrainian army.
