Look Again at the Syntax: Relational Graph Convolutional Network for Gendered Ambiguous Pronoun Resolution

Original Paper

https://arxiv.org/abs/1905.08868

Introduction

We propose an end-to-end resolver by combining pre-trained BERT with Relational Graph Convolutional Network (R-GCN). R-GCN is used for digesting structural syntactic information and learning better task-specific embeddings. Empirical results demonstrate that, under explicit syntactic supervision and without the need to fine tune BERT, R-GCN's embeddings outperform the original BERT embeddings on the coreference task. Our work obtains the state-of-the-art results on GAP dataset, and significantly improves the snippet-context baseline F1 score from 66.9% to 80.3%. We participated in the 2019 GAP Coreference Shared Task, and our codes are available online. The overall architecture is shown below.

Dataset we have

The data set is Gendered Ambiguous Pronouns (GAP), which is a gender-balanced dataset containing 8908 coreference-labeled pairs sampled from Wikipedia. The dataset contains samples Each sample contains a small paragraph that mentions the potential subject's names later refered by a target pronoun. It also came up with two candidate names for the resolver to choose from. Columns contains:

Data Preprocessing

We use SpaCy as our syntactic denpendency parser. DGL is used to transfer each dependency tree into a graph object. This DGL graph object then can be used as the input for GCN model which is also implemented by DGL. Several graphs are grouped together as a larger DGL batch-graph object for batch training setting.