Higher-order Coreference Resolution with Coarse-to-fine Inference
This addresses coreference resolution for natural language processing, with incremental improvements in accuracy and efficiency.
The paper tackles coreference resolution by introducing a differentiable approximation to higher-order inference that refines span representations iteratively, achieving significant accuracy improvements on the English OntoNotes benchmark while being far more computationally efficient.
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.