We describe our experiments in building a system to tackle task B of the BioASQ 2019 challenge on semantic question answering. We built separate systems to handle the five different types of questions in the dataset. We explored using transformer-based models using both ELMo, BERT and BioBERT. For the yesno questions, the results of our submissions using BERT ranked first in batches 3 and 4, while second best in batch 5.
Transformer Models for Question Answering at BioASQ 2019
Resta, Michele;Fagnani, Alessandro;Attardi, Giuseppe
2019-01-01
Abstract
We describe our experiments in building a system to tackle task B of the BioASQ 2019 challenge on semantic question answering. We built separate systems to handle the five different types of questions in the dataset. We explored using transformer-based models using both ELMo, BERT and BioBERT. For the yesno questions, the results of our submissions using BERT ranked first in batches 3 and 4, while second best in batch 5.File in questo prodotto:
Non ci sono file associati a questo prodotto.
I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
