РАЗРАБОТКА СЕРВИСА АНАЛИЗА ПЕРЕПИСОК С ИСПОЛЬЗОВАНИЕМ ТЕХНОЛОГИЙ ИСКУССТВЕННОГО ИНТЕЛЛЕКТА И ВЕКТОРНОЙ БАЗЫ ДАННЫХ ДЛЯ ЦИФРОВОЙ КРИМИНАЛИСТИКИ

Abstract

This paper presents an innovative correspondence analysis system for digital forensics. The service utilizes artificial intelligence technologies, including transformer-based models and vector databases, enabling semantic analysis of text messages regardless of wording or language. Integrated automatic translation and detection of unsafe content significantly enhance the capabilities of forensic analysts. The proposed message retrieval method leverages vector text representations, enabling context-based searches beyond mere keywords. Additionally, the system supports filtering by metadata such as sender, recipient, timestamps, geolocation, and message status, substantially increasing forensic accuracy and efficiency. Experiments conducted on a synthetic dataset containing 7448 messages confirmed the accuracy of message retrieval and identification of unsafe content, demonstrating robust multilingual performance. An integrated data visualization module highlights communication trends, simplifying data interpretation during investigations. The developed system offers scalability, semantic search, multilingual interaction, and unsafe content detection, suitable for law enforcement and other analytical domains.