Artificial intelligence based framework for cognitive domain analysis

Read full text Summarise with VIRAT
Authors: Mario Aragonés Lozano, Israel Pérez-Llopis, Antonio José Caballero Morcillo, Antonio León Fernández
Cited by: 0
References: 42

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

Understanding how populations perceive, interpret, and respond to unfolding events requires far more than aggregating sentiment scores or opinion labels. It demands structured, multi-dimensional assessment of how information environments shape belief, decision-making, and behaviour across political, social, cultural, and operational dimensions. This work proposes a full-featured machine learning-enabled framework for conducting such cognitive domain analysis based on open-source intelligence data drawn primarily from the cyber domain. Unlike prior systems that address isolated analytical sub-tasks, the proposed framework integrates these as atomic building blocks within higher-order analytical workflows, most notably the PMESII-PT Information Environment Assessment and the COM-B behavioural model, which serve as structured lenses for operationalising cognitive domain assessment. The solution spans data collection, multi-modal correlation, large language model-driven analysis, synthetic cognitive product generation, and interactive visualisation, making it suitable for the complete decision-support chain from data analyst to executive officer. The contribution lies not in new machine learning algorithms but in (i) a formal operationalisation of cognitive domain analysis that unifies affective, narrative, behavioural, and contextual analytical layers; (ii) an architecture that orchestrates established behavioural and information-environment models as large language model-orchestration schemas; and (iii) a dual-validation methodology that systematically quantifies the performance gap between curated benchmarks and operational open-source intelligence conditions. Validation results demonstrate time savings of 64.35–68.5% across critical intelligence analysis tasks while maintaining user acceptance ratings between 3.4 and 4.1 on a 5-point scale. Performance degradation of 15%–38% under realistic open-source intelligence multimedia conditions is documented, establishing baseline expectations for operational artificial intelligence deployment.

Fields of Study & Subjects

Sentiment Analysis and Opinion Mining Intelligence Security War Strategy Computational and Text Analysis Methods

Publication details

Journal / source
Engineering Applications of Artificial Intelligence
Volume
175
Pages
114647–114647
Publication date
28 March 2026
Language
EN