Por favor, use este identificador para citar o enlazar este ítem:
http://infotec.repositorioinstitucional.mx/jspui/handle/1027/808| INFOTEC-NLP at SemEval-2025 Task 11: A Case Study on Transformer-Based Models and Bag of Words | |
| Emmanuel Santos Rodríguez Mario Graff | |
| Acceso Abierto | |
| Atribución-NoComercial-CompartirIgual | |
| Detección de emociones Lingüística computacional | |
| Emotion detection in text is a key task in computational linguistics, challenged by linguistic ambiguities, cultural differences, and the scarcity of non-English resources, limiting its multilingual applicability. While pre-trained transformers and neural networks have shown strong performance, research remains largely English-centric, highlighting the need for inclusive, cross-linguistic approaches. This work tackles SemEval-2025 Task 11, Track A: Multilabel Emotion Detection (Muhammad et al., 2025b), predicting perceived emotions (joy, sadness, fear, anger, surprise or disgust) in text snippets. We propose a hybrid model combining XLM-RoBERTa embeddings with Bi-LSTM and multi-head attention, enhancing contextual understanding and classification across languages. Experiments on the task dataset show our model effectively captures emotional nuances, outperforming the baselines in most languages. Results show that our method improves macro-F1 scores for multilingual emotion classification. These findings highlight the value of combining transformerbased embeddings with structured sequence modeling to better represent linguistic and cultural diversity. | |
| INFOTEC Centro de Investigación e Innovación en Tecnologías de la Información y Comunicación | |
| 2025 | |
| Artículo | |
| Inglés | |
| Santos Rodríguez, E. y Mario Graff. (2025). INFOTEC-NLP at SemEval-2025 Task 11: A Case Study on Transformer-Based Models and Bag of Words. INFOTEC Centro de Investigación e Innovación en Tecnologías de la Información y Comunicación, Ciudad de México. | |
| TECNOLOGÍA DE LAS TELECOMUNICACIONES | |
| Versión publicada | |
| publishedVersion - Versión publicada | |
| Aparece en las colecciones: | Maestría en Ciencia de Datos e Información |
Cargar archivos:
| Fichero | Tamaño | Formato | |
|---|---|---|---|
| INFOTEC_MCDI_ESR_24062025.pdf | 738.55 kB | Adobe PDF | Visualizar/Abrir |