Developing of an Anti-Fake Service of Educational Purposes for the Analysis of Russian-Language News Texts

Purpose of the research. The relevance of this study is driven by the steady growth of disinformation in the digital environment, which makes the development of automated content verification systems a critical tool for ensuring national information security. In addition, such systems serve as important digital technologies for educational purposes, helping to improve the digital literacy of modern users. The aim of this paper is to develop and substantiate the architecture of an anti-fake service as a specialized digital tool for education. The study addresses the following objectives: a comparative assessment of the effectiveness of neural network models for analyzing news in Russian and the development of methodological recommendations for integrating digital services into the educational process to foster a critical attitude toward information in the contemporary Russian-language information space. Materials and methods. The anti-fake service is based on an ensemble method for training neural network models, combining recurrent neural networks (including long short-term memory architecture) and convolutional neural networks. The training was conducted on the public data set of Russian-language news “Fake and real Russian news”. A specialized preprocessing pipeline, including lemmatization and stop word removal, was implemented for Russian-language text. The quality of binary classification for the Russian-language news was assessed using a set of metrics: Accuracy, Precision, Recall, and F1-score. The digital service architecture is implemented using a microservices approach based on the FastAPI/ React technology stack, ensuring system scalability and flexibility. The results. It is developed a working prototype of an intelligent antifake service for analyzing Russian-language news texts. A key feature of the proposed digital solution is that it allows users to interactively select and combine deep learning models in real time via an intuitive web interface. This functionality not only enables news classification as reliable or fake, but also clearly demonstrates the inner workings of deep learning algorithms in an educational context, helping users to develop practical information verification skills. Experimental data confirm the effectiveness of the proposed approach to developing an anti-fake service: the highest accuracy (Accuracy 93.3%) was achieved using a convolutional neural network, owing to its ability to identify local semantic patterns in texts. The use of a probabilistic ensemble further improved the reliability and robustness of binary classification of Russian-language news. Conclusion. The conducted research solves a pressing problem at the intersection of information security and pedagogy countering disinformation through the development of digital literacy. The novelty and key advantage of the developed anti-fake service lie in its functional duality: it serves both as a tool for the automatic binary classification of Russian-language news texts and at the same time, as a service for developing users’ practical information verification skills through interactive engagement with neural network models. The anti-fake service has significant didactic potential and can be integrated into educational programs for courses to develop students’ digital literacy, including its use in courses on computer science, information security, machine learning, and other disciplines. Thus, the developed anti-fake service functions both as a tool for automatic verification of Russian-language content and as a digital educational tool that promotes the formation of a critical attitude toward information in the digital media environment.