Sequence to Sequence Coreference Resolution

Gorka Urbizu, Ander Soraluze, Olatz Arregi


Abstract
Until recently, coreference resolution has been a critical task on the pipeline of any NLP task involving deep language understanding, such as machine translation, chatbots, summarization or sentiment analysis. However, nowadays, those end tasks are learned end-to-end by deep neural networks without adding any explicit knowledge about coreference. Thus, coreference resolution is used less in the training of other NLP tasks or trending pretrained language models. In this paper we present a new approach to face coreference resolution as a sequence to sequence task based on the Transformer architecture. This approach is simple and universal, compatible with any language or dataset (regardless of singletons) and easier to integrate with current language models architectures. We test it on the ARRAU corpus, where we get 65.6 F1 CoNLL. We see this approach not as a final goal, but a means to pretrain sequence to sequence language models (T5) on coreference resolution.
Anthology ID:
2020.crac-1.5
Volume:
Proceedings of the Third Workshop on Computational Models of Reference, Anaphora and Coreference
Month:
December
Year:
2020
Address:
Barcelona, Spain (online)
Venues:
COLING | CRAC
SIG:
Publisher:
Association for Computational Linguistics
Note:
Pages:
39–46
Language:
URL:
https://aclanthology.org/2020.crac-1.5
DOI:
Bibkey:
Cite (ACL):
Gorka Urbizu, Ander Soraluze, and Olatz Arregi. 2020. Sequence to Sequence Coreference Resolution. In Proceedings of the Third Workshop on Computational Models of Reference, Anaphora and Coreference, pages 39–46, Barcelona, Spain (online). Association for Computational Linguistics.
Cite (Informal):
Sequence to Sequence Coreference Resolution (Urbizu et al., CRAC 2020)
Copy Citation:
PDF:
https://aclanthology.org/2020.crac-1.5.pdf
Code
 gorka96/text2cor