An ELECTRA-Based Model for Neural Coreference Resolution
Metadatos
Mostrar el registro completo del ítemEditorial
IEEE
Materia
Conference resolution ELECTRA Neural language model OntoNotes Natural language processing
Fecha
2022-07-12Referencia bibliográfica
F. Gargiulo... [et al.], "An ELECTRA-Based Model for Neural Coreference Resolution," in IEEE Access, vol. 10, pp. 75144-75157, 2022, doi: [10.1109/ACCESS.2022.3189956]
Resumen
In last years, coreference resolution has received a sensibly performance boost exploiting
different pre-trained Neural Language Models, from BERT to SpanBERT until Longformer. This work is
aimed at assessing, for the rst time, the impact of ELECTRA model on this task, moved by the experimental
evidence of an improved contextual representation and better performance on different downstream tasks.
In particular, ELECTRA has been employed as representation layer in an assessed neural coreference
architecture able to determine entity mentions among spans of text and to best cluster them. The architecture
itself has been optimized: i) by simplifying the modality of representation of spans of text but still considering
both the context they appear and their entire content, ii) by maximizing both the number and length
of input textual segments to exploit better the improved contextual representation power of ELECTRA,
iii) by maximizing the number of spans of text to be processed, since potentially representing mentions,
preserving computational ef ciency. Experimental results on the OntoNotes dataset have shown the effectiveness
of this solution from both a quantitative and qualitative perspective, and also with respect to other
state-of-the-art models, thanks to a more pro cient token and span representation. The results also hint at
the possible use of this solution also for low-resource languages, simply requiring a pre-trained version of
ELECTRA instead of language-speci c models trained to handle either spans of text or long documents.