COVID-19 Spanish Sentiment Polarity Analyzer (2025)
Multimodal model (BETO + Vision Transformer) for polarity analysis of Spanish social-media posts about COVID-19, reaching a 96% F1 score.
Research & Projects
Mexican Boxing Sentiment Polarity Analyzer (2025)
Multimodal model (BETO + Vision Transformer) analyzing Spanish social-media sentiment around Saúl «Canelo» Álvarez’s fights, reaching a 77.76% F1 score.
Text Detection in Images (2024)
A YOLOv8 / Ultralytics model to detect text embedded in images, such as memes in social-media posts.
Phishing Detection on Emails (2023)
New phishing-detection models for email, including a multilabel model that detects persuasion principles in English with LLMs.
#8M on Twitter (2023)
Social-network analysis of the 2023 International Women’s Day movement on Twitter: community detection, temporal analysis, sentiment, and user classification.
Human Rights Recommendations Labeling Automation (2023)
Multilabel models to automatically classify human-rights recommendations from intergovernmental organizations using LLMs.
Opinion Mining for Electoral Preference Modeling (2023)
Opinion mining on Twitter (now X) via the Twitter API to model electoral preferences.
Proyecto Imagen de México (2022–2023)
Analyzing Mexico’s image abroad from news and social-media sources via APIs; fine-tuned LLMs improved F1 by ~14% over previous work.
Impact of Mining in Latin America (2022)
Collection and analysis of mining-activity reports from alternative digital sources across Latin America (2010–2022).