References & Citations
Computer Science > Computation and Language
Title: Fine-grained Information Status Classification Using Discourse Context-Aware BERT
(Submitted on 26 Oct 2020 (v1), last revised 1 Nov 2020 (this version, v2))
Abstract: Previous work on bridging anaphora recognition (Hou et al., 2013a) casts the problem as a subtask of learning fine-grained information status (IS). However, these systems heavily depend on many hand-crafted linguistic features. In this paper, we propose a simple discourse context-aware BERT model for fine-grained IS classification. On the ISNotes corpus (Markert et al., 2012), our model achieves new state-of-the-art performance on fine-grained IS classification, obtaining a 4.8 absolute overall accuracy improvement compared to Hou et al. (2013a). More importantly, we also show an improvement of 10.5 F1 points for bridging anaphora recognition without using any complex hand-crafted semantic features designed for capturing the bridging phenomenon. We further analyze the trained model and find that the most attended signals for each IS category correspond well to linguistic notions of information status.
Submission history
From: Yufang Hou [view email][v1] Mon, 26 Oct 2020 22:30:17 GMT (68kb,D)
[v2] Sun, 1 Nov 2020 14:36:49 GMT (68kb,D)
Link back to: arXiv, form interface, contact.