Compiling a Corpus of User-Generated Content Units for the Detection of Social Problems
Rocío Jiménez Briones · Revista Electrónica de Lingüística Aplicada · 2025
Social media platforms like Facebook, X, and Instagram provide valuable information about daily and global social problems through their user-generated content units. Such platforms turn their users into social sensors capable of identifying problems such as violence against women or natural hazards. Within the field of crowdsensing, which aims to extract useful information from these social sensors, we introduce a proposal for identifying problems in ALLEGRO (http://allegro.ucam.edu/). Based on previous studies within this smart multimodal system, our research first explains the conceptual framework for addressing one of these problems, namely, elderly institutional abuse, within the text analysis module of ALLEGRO or DIAPASON. We also detail the methodology and challenges of compiling a subcorpus of tweets related to this problem. Such a specific subcorpus will contribute to ALLEGRO’s comprehensive corpus of social problems, which is being built as a training set for deep learning models in text classification.