Incorporating Structural Information for Better Coreference Resolution

Kong, Fang; Fu, Jian

doi:10.24963/ijcai.2019/700

“…Avg. F1 (Clark and Manning, 2016b) 73.85 65.42 69.38 67.53 56.41 61.47 62.84 57.62 60.12 63.66 (Clark and Manning, 2016a) 73.64 65.62 69.40 67.48 56.94 61.76 62.46 58.60 60.47 63.88 (Kong and Fu, 2019) 76…”

Section: Datasetsmentioning

confidence: 99%

Tackling Zero Pronoun Resolution and Non-Zero Coreference Resolution Jointly

Chen¹,

Gu²,

Qu³

et al. 2021

Proceedings of the 25th Conference on Computational Natural Language Learning

View full text Add to dashboard Cite

Zero pronoun resolution aims at recognizing dropped pronouns and pointing out their anaphoric mentions, while non-zero coreference resolution targets at clustering mentions referring to the same entity. Existing efforts often deal with the two problems separately regardless of their close essential correlations. In this paper, we investigate the possibility of jointly solving zero pronoun resolution and coreference resolution via a novel end-to-end neural model. Specifically, we design a gapmasked self-attention model that encodes gaps and tokens in the same space, where gaps could capture valuable contextual information according to their surrounding tokens while tokens could maintain original sequential information without disturbance. Additionally, we also propose a two-stage interaction mechanism to make full use of the exclusive relationship between zero pronouns and mentions. Our empirical study conducted on the OntoNotes 5.0 Chinese dataset shows that our model could outperform corresponding stateof-the-art approaches on both tasks.

show abstract

“…In order to effectively incorporate constituent syntax structures, we propose a graph-based neural coreference model. Previous works (Trieu et al 2019;Kong and Jian 2019) introduced constituent trees as hard constraints to filter invalid mentions, which helps obtain better mention detectors and overall coreference resolvers. However, these methods do not preserve the full structure of original trees, as a result of their simplification of the tree as node traversal sequences with a collection of path features, or by ignoring the hierarchical structure entirely.…”

Section: Introductionmentioning

confidence: 99%

Incorporating Constituent Syntax for Coreference Resolution

Fan¹,

Cohn²

2022

Preprint

1

0

View full text Add to dashboard Cite

Syntax has been shown to benefit Coreference Resolution from incorporating long-range dependencies and structured information captured by syntax trees, either in traditional statistical machine learning based systems or recently proposed neural models. However, most leading systems use only dependency trees. We argue that constituent trees also encode important information, such as explicit span-boundary signals captured by nested multi-word phrases, extra linguistic labels and hierarchical structures useful for detecting anaphora. In this work, we propose a simple yet effective graph-based method to incorporate constituent syntactic structures. Moreover, we also explore to utilise higher-order neighbourhood information to encode rich structures in constituent trees. A novel message propagation mechanism is therefore proposed to enable information flow among elements in syntax trees. Experiments on the English and Chinese portions of OntoNotes 5.0 benchmark show that our proposed model either beats a strong baseline or achieves new state-of-the-art performance. 1

show abstract

“…In order to effectively incorporate constituent syntax structures, we propose a graph-based neural coreference model. Previous works (Trieu et al 2019;Kong and Jian 2019) introduced constituent trees as hard constraints to filter invalid mentions, which helps obtain better mention detectors and overall coreference resolvers. However, these methods do not preserve the full structure of original trees, as a result of their…”

Section: Introductionmentioning

confidence: 99%

Incorporating Constituent Syntax for Coreference Resolution

Jiang

¹

,

Cohn

²

2022

AAAI

2

0

View full text Add to dashboard Cite

Syntax has been shown to benefit Coreference Resolution from incorporating long-range dependencies and structured information captured by syntax trees, either in traditional statistical machine learning based systems or recently proposed neural models. However, most leading systems use only dependency trees. We argue that constituent trees also encode important information, such as explicit span-boundary signals captured by nested multi-word phrases, extra linguistic labels and hierarchical structures useful for detecting anaphora. In this work, we propose a simple yet effective graph-based method to incorporate constituent syntactic structures. Moreover, we also explore to utilise higher-order neighbourhood information to encode rich structures in constituent trees. A novel message propagation mechanism is therefore proposed to enable information flow among elements in syntax trees. Experiments on the English and Chinese portions of OntoNotes 5.0 benchmark show that our proposed model either beats a strong baseline or achieves new state-of-the-art performance. Code is available at https://github.com/Fantabulous-J/Coref-Constituent-Graph.

show abstract

Incorporating Structural Information for Better Coreference Resolution

Cited by 11 publications

References 15 publications

Tackling Zero Pronoun Resolution and Non-Zero Coreference Resolution Jointly

Tackling Zero Pronoun Resolution and Non-Zero Coreference Resolution Jointly

Incorporating Constituent Syntax for Coreference Resolution

Incorporating Constituent Syntax for Coreference Resolution

Contact Info

Product

Resources

About