2026-10-11T05:21:58.075Z http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai

ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai:3:12177 2026-09-18T18:09:54Z 3

Cloud System Logs Analysis With NLP and XGBoost Optimized by Adapted Sine Cosine Algorithm, Chapter in LNNS Lecture Notes in Networks and Systems: ICITI 2025: International Conference on Information Technology and Intelligence, Springer, volume 2031 2026 http://ezaposleni.singidunum.ac.rs/rest/sciNaucniRezultati/oai/record/3/12177 https://link.springer.com/chapter/10.1007/978-3-032-29319-0_2 J. Maricic S. Anetic T. Zivkovic V. Thomas M. Zivkovic B. Radomirovic L. Anicin N. Bacanin Effective log monitoring is essential for securing the reliability and performance levels of large-scale cloud-based environments. As these systems grow in complexity, traditional monitoring methods struggle to keep pace, often resulting in excessive computational demands and reduced effectiveness. This research explores a hybrid approach that combines natural language processing techniques with an advanced XGBoost classification model to identify anomalies in logs generated by cloud systems. Since the classification performance is heavily relying on well-tuned hyperparameters, the framework incorporates a customized version of the renowned sine cosine algorithm for refining the tuning process. The optimized XGBoost models were evaluated on publicly available, real-world log datasets. Experimental results demonstrate exceptional performance, with leading models achieving accuracy rates around 98.53%, underpinning the framework’s potential as a scalable and intelligent solution for anomaly detection in cloud infrastructure logging. bookPart Springer, Cham 15 28 10.1007/978-3-032-29319-0_2 LNNS Lecture Notes in Networks and Systems: ICITI 2025: Proceedings of International Conference on Information Technology and Intelligence, volume 2031