Overview of ASQP-PT at IberLEF 2025: Overview of the Task on Aspect-Sentiment Quadruple Prediction in Portuguese

Resumen

This paper reports the results of ASQP-PT 2025, the competition on Aspect-Sentiment Quadruple Prediction in Portuguese at IberLEF 2025. The ABCD Team participated in all four tasks, using an end-to-end approach based on encoderdecoder transformer models, fine-tuned for this task. The VerbaNexAI team participated in the Aspect Extraction task, with an approach based on fine-tuning a transformer encoder model as a sequence labeling problem. The ABCD Team obtained 0.46 of F-Measure in the ASQP task, surpassing the competition baseline of 0.40 F-Measure.