article · Information Processing & Management
With the increasing frequency and sophistication of cyberattacks, cybersecurity has become a strategic priority, creating an urgent demand for a skilled workforce. Online job ads in cyberspace are a rich source of information related to the changing dynamics of labor markets, sought-after skills, and employer expectations. However, available methods for extracting relevant information from job advertisements, including for cybersecurity careers, have thus far relied on either manual or semi-automated approaches that, in general, are not scalable, are prone to error, and require prior knowledge and expertise in natural language processing (NLP). This study presents a new, fully automated framework for cybersecurity labor-market analysis with potential for application in other knowledge-intensive industries. The fundamental innovation is the use of retrieval-augmented generation (RAG), powered by large language models (GPT-3.5, GPT-4.1, and Meta Llama 3) to turn unstructured job postings into a structured and analyzable format. This methodology is applied to real job postings from LinkedIn, Indeed, and Rekrute, with a particular focus on Morocco’s cybersecurity sector. From a balanced dataset of 27,360 job advertisements, this study analyzes 1681 cybersecurity roles. Results demonstrate the effectiveness of the proposed approach: GPT-4.1, used as the master evaluator, achieves the highest scores across the six metrics: 96.276% (Correctness), 100% (Completeness), 99.952% (Relevance), 100% (Format Adherence), 99.762% (Language Consistency), and 99.881% (Clarity). Moreover, this study shows that most job postings target mid-level candidates with three or more years of experience and advanced degrees, primarily master’s or engineering degrees. CISSP is the most sought-after certification. By automating and scaling the analysis of labor market data, this research contributes to a broader societal objective of strengthening the cybersecurity workforce and fostering secure digital societies.
This page summarises published work. The authoritative version sits with the publisher.
DOI: 10.1016/j.ipm.2026.105067
Is something wrong with this record? Report it or request removal.
Discussion
Have you built on this work, tried to replicate it, or seen it applied in practice? Share what you know. Verified researchers and MARATTO™ domain experts can open a discussion, and any member can reply. Contributions are reviewed before they appear.
No discussion yet. Open the first thread.
New to MARATTO™? Create a free account.