University of Southampton
Log-to-graph methodology for malicious activities detection in automated cloud infrastructure
Abstract
dc:description.abstractCloud computing platforms increasingly rely on automated orchestration, in which security-relevant behaviour is primarily captured in operational logs rather than infixed network- or host-level indicators. Detecting malicious activities in these environments remains challenging because traditional statistical and sequence-based approaches often fail to capture complex behavioural dependencies across services, and many graph-based methods rely on extensive labelled data or offer limited interpretability. This thesis presents an end-to-end semi-supervised methodology for detecting malicious activities in automated cloud infrastructures using graph representations derived from operational logs. The proposed pipeline transforms heterogeneous logs into semantically enriched session graphs that preserve both operational semantics and behavioural structure. By integrating semantic, structural, and latent graph characteristics, the methodology employs a multi-view ensemble detection framework that supports both semi-supervised and fully label-free operation, enabling robust anomaly detection across diverse cloud environments. The methodology is evaluated on four public datasets: CTDD, HDFS, Thunderbird, and Hadoop. Results demonstrate strong and consistent performance, achieving anF1-score of 98.97% and a macro-average F1-score of 99.03±0.33% on CTDD across five random seeds, while remaining competitive or superior to established baselines. Overall, his thesis contributes an interpretable and scalable approach for transforming operational cloud logs into semantically structured graph representations and for detecting malicious workflows, providing behavioural insights that can support cloud security monitoring and analyst decision-making.<br/>
Degree
thesis:*- Name dc:type.qualificationname
- Ph.D.
- Level dc:type.qualificationlevel
- doctoral
- Grantor dc:publisher.institution
- University of Southampton
- Year dc:date.issued
- 2026
Author and committee
dc:creator, dc:contributor.*- Author dc:creator
-
- Alaskar, Mada Mansour A
- Advisors dc:contributor.advisor
-
- Al Hashimy, Nawfal
- Atamli, Ahmad
- Hettrick, Simon
- ORCID person:identifier.orcid
Rights
- Language dc:language
- English, eng
Identifiers
dc:identifier.*- DOI dc:identifier.doi
- https://doi.org/10.5258/SOTON/PG/T077