Generative AI-based tool for efficient processing of Human Rights violations-related news
Raúl Ernesto Guillén Hernández, Ronaldo Armando Canizales Turcios, Mauricio Erazo · 2024
The increasing volume of online news and reports related to human rights abuses presents a significant challenge for organizations and policymakers, as traditional manual monitoring methods are often overwhelmed, leading to delays and gaps in coverage. This paper introduces a fast, accurate, and scalable approach for web scraping and processing such news, leveraging NLP techniques and open-source LLMs. Our method significantly accelerates data documentation and systematization, supporting evidence-based decision-making and enhancing real-time monitoring of human rights violations. Tested on a comprehensive dataset of 450 news from three popular websites, our approach achieves results in just 20, 10, and 6.4 seconds—tasks that would otherwise take hours manually—demonstrating its efficiency and potential for widespread application.