Evaluating the Impact of a Hierarchical Discourse Representation on Entity Coreference Resolution Performance
Recent work on entity coreference resolution (CR) follows current trends in\nDeep Learning applied to embeddings and relatively simple task-related\nfeatures. SOTA models do not make use of hierarchical representations of\ndiscourse structure. In this work, we leverage automatically constructed\ndiscourse parse trees within a neural approach and demonstrate a significant\nimprovement on two benchmark entity coreference-resolution datasets. We explore\nhow the impact varies depending upon the type of mention.\n