We introduce a fully differentiable approximation to higher-order inference
for coreference resolution. Our approach uses the antecedent distribution from
a span-ranking architecture as an attention mechanism to iteratively refine
span representations. This enables the model to softly consider multiple hops
in the predicted clusters. To alleviate the computational cost of this
iterative process, we introduce a coarse-to-fine approach that incorporates a
less accurate but more efficient bilinear factor, enabling more aggressive
pruning without hurting accuracy. Compared to the existing state-of-the-art
span-ranking approach, our model significantly improves accuracy on the English
OntoNotes benchmark, while being far more computationally efficient.

引入一种全可微分的高阶推理近似方法，用于指代消解，该方法使用跨度排名架构中的前先分布作为注意机制来迭代地改进跨度表示，从而允许模型对预测聚类中的多个跳进行软考虑，同时引入了粗到细的方法来降低计算成本。与现有的跨度排名方法相比，我们的模型在英语 OntoNotes 基准测试上显著提高了准确性，同时更加计算效率。