International Journal of Scientific Research and Engineering Development

Modern distributed IT systems generate enormous volumes of operational data every second. Server logs, application event streams, and infrastructure monitoring feeds collectively encode the real-time health of any enterprise platform. Yet the vast majority of operations teams still rely on threshold-based alerting rules that fire only after a problem has already become visible, by which point substantial damage in terms of downtime, user impact, or data integrity may already have occurred. The challenge is not a shortage of data; it is a shortage of context.
This paper presents a Context-Aware AIOps framework designed to address that gap by detecting hidden anomalies across multi-source system logs before they escalate into incidents. The framework ingests heterogeneous log streams from server infrastructure, application runtimes, and monitoring exporters, correlates them through a shared semantic model, and applies a combination of Isolation Forest-based outlier detection and Autoencoder-driven reconstruction error analysis to identify behavioral patterns that deviate from learned baselines. A context enrichment layer associates raw anomaly signals with service topology, temporal patterns, and historical incident fingerprints to suppress false positives and surface only those deviations that warrant operator attention.
We implemented the framework using Python, Pandas, Scikit-learn, and a Flask-based monitoring dashboard, and evaluated it on a synthetic dataset of 120,000 log events generated to reflect realistic enterprise workloads. The system achieved an anomaly detection accuracy of 94.3%, a precision of 92.7%, and a recall of 95.1%, outperforming standalone threshold-based detection by a margin of 21 percentage points in F1 score. The results suggest that contextaware multi-source log correlation is a practical and effective approach for improving anomaly detection in production AIOps deployments.

R.Ramakrishnan, Keerthana N,"Context-Aware AIOps Framework for Detecting Hidden Anomalies Using Multi-Source System Logs" International Journal of Scientific Research and Engineering Development, V9(3): Page(1473-1483) May-June 2026. ISSN: 2581-7175. www.ijsred.com. Published by Scientific and Academic Research Publishing.