Integration of GenAI and Behavioral Analysis in an Authentication System
Computer Science and Interdisciplinary Research Journal · 2025
DOI: 10.70862/csir.2025.0202-17 ↗Kurz gefasst
Fachartikel in der Zeitschrift Computer Science and Interdisciplinary Research Journal, 2025. Verfasst von Georgi Atanasov und 2 weiteren Autor:innen. Der Volltext ist frei zugänglich (Open Access, CC BY-NC-ND 4.0).
Abstract (Originalsprache)
This paper presents a model of a security-enhanced web authentication system that combines behavioral analysis, generative artificial intelligence, and an event-driven architecture. The aim of the study is to develop a dynamic monitoring and alerting mechanism capable of identifying unusual activity based on contextual and behavioral indicators, without imposing additional authentication steps on the user. The system leverages Redis with Pub/Sub and TTL notifications for reactive event processing, generative AI for automatic summarization of logs into a human-readable format, and the Isolation Forest algorithm for experimental detection of behavioral anomalies. The results demonstrate that the proposed architecture can effectively identify suspicious sessions, provide interpretable textual explanations of user actions, and react to key events in real time while maintaining low latency.
Zitieren
APA
Atanasov, G., Yotov, K., & Hadzhikolev, E. (2025). Integration of GenAI and Behavioral Analysis in an Authentication System. Computer Science and Interdisciplinary Research Journal. https://doi.org/10.70862/csir.2025.0202-17
Vancouver
Atanasov G, Yotov K, Hadzhikolev E. Integration of GenAI and Behavioral Analysis in an Authentication System. Computer Science and Interdisciplinary Research Journal. 2025. doi:10.70862/csir.2025.0202-17
BibTeX
@article{Atanasov2025,
author = {Atanasov, Georgi and Yotov, Kostadin and Hadzhikolev, Emil},
title = {Integration of GenAI and Behavioral Analysis in an Authentication System},
journal = {Computer Science and Interdisciplinary Research Journal},
year = {2025},
doi = {10.70862/csir.2025.0202-17}
}
EUVIMED