Unmasking Media Bias, Economic Resilience, and the Hidden Patterns of Global Catastrophes DOI Open Access
Fahim Sufi, Musleh Alsulami

Sustainability, Journal Year: 2025, Volume and Issue: 17(9), P. 3951 - 3951

Published: April 28, 2025

The increasing frequency and destructiveness of natural disasters necessitate scalable, transparent, timely analytical frameworks for risk reduction. Traditional disaster datasets—curated by intergovernmental bodies such as EM-DAT UNDRR—face limitations in spatial granularity, temporal responsiveness, accessibility. This study addresses these introducing a novel, AI-enhanced intelligence framework that leverages 19,130 publicly available news articles from 453 global sources between September 2023 March 2025. Using OpenAI’s GPT-3.5 Turbo model classification metadata extraction, the transforms unstructured text into structured variables across five key dimensions: severity, location, media coverage, economic resilience, casualties. Hypotheses were tested using statistical modeling, geospatial aggregation, time series analysis. Findings confirm modest but significant correlation severity casualties (ρ=0.12, p<10−60), stronger average regional impact (ρ=0.31, p<10−10). Media amplification bias was empirically demonstrated: hurricanes received most coverage (5599 articles), while under-reported earthquakes accounted over 3 million deaths. Economic resilience showed statistically weak protective effect on fatalities (β=−0.024, p=0.041). Disaster increased substantially (slope η1=53.17, R2=0.32), though remained stable. GPT-based achieved high F1-score (0.91), demonstrating robust semantic accuracy, not mortality prediction. validates feasibility AI-curated, open-access data empirical hypothesis testing science, offering sustainable alternative to closed datasets enabling real-time policy feedback loops, particularly vulnerable, data-scarce regions.

Language: Английский

Mathematical Modeling and Clustering Framework for Cyber Threat Analysis Across Industries DOI Creative Commons
Fahim Sufi, Musleh Alsulami

Mathematics, Journal Year: 2025, Volume and Issue: 13(4), P. 655 - 655

Published: Feb. 17, 2025

The escalating prevalence of cyber threats across industries underscores the urgent need for robust analytical frameworks to understand their clustering, prevalence, and distribution. This study addresses challenge quantifying analyzing relationships between 95 distinct cyberattack types 29 industry sectors, leveraging a dataset 9261 entries filtered from over 1 million news articles. Existing approaches often fail capture nuanced patterns such complex datasets, justifying innovative methodologies. We present rigorous mathematical framework integrating chi-square tests, Bayesian inference, Gaussian Mixture Models (GMMs), Spectral Clustering. identifies key patterns, as 1150 Zero-Day Exploits clustered in IT Telecommunications sector, 732 Advanced Persistent Threats (APTs) Government Public Administration, Malware with posterior probability 0.287 dominating Healthcare sector. Temporal analyses reveal periodic spikes, Exploits, persistent presence Social Engineering Attacks, 1397 occurrences industries. These findings are quantified using significance scores (mean: 3.25 ± 0.7) probabilities, providing evidence industry-specific vulnerabilities. research offers actionable insights policymakers, cybersecurity professionals, organizational decision makers by equipping them data-driven understanding sector-specific risks. formulations replicable scalable, enabling organizations allocate resources effectively develop proactive defenses against emerging threats. By bridging theory real-world challenges, this delivers impactful contributions toward safeguarding critical infrastructure digital assets.

Language: Английский

Citations

1
Enhancing News Articles: Automatic SEO Linked Data Injection for Semantic Web Integration DOI Creative Commons
Hamza Salem, Hadi Salloum,

Osama Orabi

et al.

Applied Sciences, Journal Year: 2025, Volume and Issue: 15(3), P. 1262 - 1262

Published: Jan. 26, 2025

This paper presents a novel solution aimed at enhancing news web pages for seamless integration into the Semantic Web. By utilizing advanced pattern mining techniques alongside OpenAI’s GPT-3, we rewrite articles to improve their readability and accessibility Google News aggregators. Our approach is characterized by its methodological rigour evaluated through quantitative metrics, validated using Google’s Rich Results Test API confirm adherence structured data guidelines. In this process, “Pass” in taken as an indication of eligibility rich results, demonstrating effectiveness our generated data. The impact work threefold: it advances technological substantial segment Web, promotes adoption Web technologies within sector, significantly enhances discoverability aggregator platforms. Furthermore, facilitates broader dissemination content diverse audiences. submission introduces innovative substantiated empirical evidence soundness, thereby making significant contribution field research, particularly context media articles.

Language: Английский

Citations

0
Unmasking Media Bias, Economic Resilience, and the Hidden Patterns of Global Catastrophes DOI Open Access
Fahim Sufi, Musleh Alsulami

Sustainability, Journal Year: 2025, Volume and Issue: 17(9), P. 3951 - 3951

Published: April 28, 2025

The increasing frequency and destructiveness of natural disasters necessitate scalable, transparent, timely analytical frameworks for risk reduction. Traditional disaster datasets—curated by intergovernmental bodies such as EM-DAT UNDRR—face limitations in spatial granularity, temporal responsiveness, accessibility. This study addresses these introducing a novel, AI-enhanced intelligence framework that leverages 19,130 publicly available news articles from 453 global sources between September 2023 March 2025. Using OpenAI’s GPT-3.5 Turbo model classification metadata extraction, the transforms unstructured text into structured variables across five key dimensions: severity, location, media coverage, economic resilience, casualties. Hypotheses were tested using statistical modeling, geospatial aggregation, time series analysis. Findings confirm modest but significant correlation severity casualties (ρ=0.12, p<10−60), stronger average regional impact (ρ=0.31, p<10−10). Media amplification bias was empirically demonstrated: hurricanes received most coverage (5599 articles), while under-reported earthquakes accounted over 3 million deaths. Economic resilience showed statistically weak protective effect on fatalities (β=−0.024, p=0.041). Disaster increased substantially (slope η1=53.17, R2=0.32), though remained stable. GPT-based achieved high F1-score (0.91), demonstrating robust semantic accuracy, not mortality prediction. validates feasibility AI-curated, open-access data empirical hypothesis testing science, offering sustainable alternative to closed datasets enabling real-time policy feedback loops, particularly vulnerable, data-scarce regions.

Language: Английский

Citations

0