Carlos E. Anchundia
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Carlos E. Anchundia has a total of 12 co-authors. Here are the top co-authors by number of shared publications:
- Patricio Zambrano
Escuela Politécnica Nacional
3 shared publications - Adrian Esteban Paguay Montenegro
Escuela Politécnica Nacional, Facultad de Ingeniería de Sistemas, Quito 170525, Ecuador
2 shared publications - Jenny Torres
Escuela Politécnica Nacional
2 shared publications - Johan Sebastian Illicachi Manzano
Escuela Politécnica Nacional, Facultad de Ingeniería de Sistemas, Quito 170525, Ecuador
2 shared publications - Andrea Damarys Oña Calahorrano
Escuela Politécnica Nacional, Facultad de Ingeniería de Sistemas, Quito 170525, Ecuador
1 shared publication - Efrain R. Fonseca C.
Universidad de las Fuerzas Armadas ESPE, Quito, Ecuador
1 shared publication
Latest Publications
- Open Access
- 16 September 2026
- Published by MDPI AG in Informatics
- Vol. 13 (9) , 151
Inappropriate Content Classification Model for Digital Violence Detection Using Hybrid Data
This study presents an inappropriate-content classification framework for digital-violence detection in Ecuadorian Spanish, addressing extreme class imbalance and dialectal variation. Data were collected in a post-API setting through a Selenium-based scraping pipeline that reconstructs conversational context using a window of ? = 3 prior interventions. An initial zero-shot labeling attempt with LLMs (Hermes) revealed severe cultural misinterpretation, overestimating the Violence class by 50 times (97.5% false alerts), which motivated full human validation and targeted data engineering. To correct imbalance without contaminating evaluation, the corpus was split before augmentation (70/15/15), and minority classes were selectively leveled via few-shot generation with LLaMA 3.1, followed by strict deduplication and cosine-similarity filtering (? = 0.85) to preserve semantic diversity. Model selection compared BETO and mBERT, with BETO outperforming. Across four training scenarios, naïve oversampling produced artificially inflated metrics indicative of overfitting, whereas the proposed cost-sensitive and regularized configuration (BETO with semantic deduplication and weighted loss) achieved 94.39% accuracy, 0.9429 weighted F1, and 0.9022 macro F1, significantly improving recovery of critical classes. Results highlight that hybrid data are effective only when carefully curated and paired with leakage-free evaluation protocols. This work demonstrates how machine learning innovation and knowledge extraction from heterogeneous data can be combined to build robust models for digital-violence detection in low-resource, culturally specific contexts.
- 16 September 2026
- Published by MDPI AG in Informatics
- Vol. 13 (9) , 151
Open AccessInappropriate Content Classification Model for Digital Violence Detection Using Hybrid Data
This study presents an inappropriate-content classification framework for digital-violence detection in Ecuadorian Spanish, addressing extreme class imbalance and dialectal variation. Data were collected in a post-API setting through a Selenium-based scraping pipeline that reconstructs conversational context using a window of ? = 3 prior interventions. An initial zero-shot labeling attempt with LLMs (Hermes) revealed severe cultural misinterpretation, overestimating the Violence class by 50 times (97.5% false alerts), which motivated full human validation and targeted data engineering. To correct imbalance without contaminating evaluation, the corpus was split before augmentation (70/15/15), and minority classes were selectively leveled via few-shot generation with LLaMA 3.1, followed by strict deduplication and cosine-similarity filtering (? = 0.85) to preserve semantic diversity. Model selection compared BETO and mBERT, with BETO outperforming. Across four training scenarios, naïve oversampling produced artificially inflated metrics indicative of overfitting, whereas the proposed cost-sensitive and regularized configuration (BETO with semantic deduplication and weighted loss) achieved 94.39% accuracy, 0.9429 weighted F1, and 0.9022 macro F1, significantly improving recovery of critical classes. Results highlight that hybrid data are effective only when carefully curated and paired with leakage-free evaluation protocols. This work demonstrates how machine learning innovation and knowledge extraction from heterogeneous data can be combined to build robust models for digital-violence detection in low-resource, culturally specific contexts.
- 2 September 2026
- Published by Springer Nature
- p. 133-145
Approach for Management Antipattern Detection on Scrum Software Projects
- 2 September 2026
- Published by Springer Nature
- p. 133-145
Approach for Management Antipattern Detection on Scrum Software Projects
- 2 September 2026
- Published by Springer Nature
- p. 297-315
A Reactive System for Inappropriate Content Moderation: An Open Source Prototype
- 2 September 2026
- Published by Springer Nature
- p. 297-315
A Reactive System for Inappropriate Content Moderation: An Open Source Prototype
- 9 August 2025
- Published by Springer Nature
- p. 237-256
Gender-Based Violence as a Cyberthreat: The Impact of Social Media on Digital Security
- 9 August 2025
- Published by Springer Nature
- p. 237-256
Gender-Based Violence as a Cyberthreat: The Impact of Social Media on Digital Security
- Open Access
- 7 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 8992-9004
Resources for Reproducibility of Experiments in Empirical Software Engineering: Topics Derived From a Secondary Study
- 7 January 2020
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 8, 8992-9004
Open AccessResources for Reproducibility of Experiments in Empirical Software Engineering: Topics Derived From a Secondary Study