Research map / Detection and segmentation / Open-set and few-shot detection
Predicting Above-Sentence Discourse Structure Using Distant Supervision from Topic Segmentation
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| Add topic-segmentation supervision from Wiki to generate discourse trees | High-level discourse tree generation · RST-DT S-D | RST Parseval micro-average precision: 66.99 -> 68.13 | Helped |
| Add topic-segmentation supervision from Wiki to generate discourse trees | High-level discourse tree generation · GUM S-D | RST Parseval micro-average precision: 64.69 -> 65.84 | Helped |
| Add topic-segmentation supervision from Wiki to generate discourse trees | High-level discourse tree generation · GUM S-P | RST Parseval micro-average precision: 73.37 -> 76.98 | Helped |
| Add topic-segmentation supervision from Wiki to generate discourse trees | High-level discourse tree generation · GUM P-D | RST Parseval micro-average precision: 69.88 -> 63.53 | Hurt |
| Add the context modeling component (coherence module and restricted self-attention) (ablation) | TS_Wiki discourse tree generation · RST-DT S-D | RST Parseval micro-average precision: 67.47 -> 68.13 | Helped |
| Add the context modeling component (coherence module and restricted self-attention) (ablation) | TS_Wiki discourse tree generation · GUM S-D | RST Parseval micro-average precision: 65.38 -> 65.84 | Helped |
| Add the context modeling component (coherence module and restricted self-attention) (ablation) | TS_Wiki discourse tree generation · RST-DT P-D | RST Parseval micro-average precision: 68.61 -> 69.78 | Helped |
| Add the context modeling component (coherence module and restricted self-attention) (ablation) | TS_Wiki discourse tree generation · GUM P-D | RST Parseval micro-average precision: 64.71 -> 63.53 | Hurt |
| Train on Wiki instead of in-domain RST-DT | Topic-segmentation-based discourse tree generation · RST-DT S-D | RST Parseval micro-average precision: 65.96 -> 68.13 | Helped |
| Fine-tune the Wiki-trained topic segmenter on RST-DT | Topic-segmentation-based discourse tree generation · RST-DT S-D | RST Parseval micro-average precision: 68.13 -> 65.84 | Hurt |
| Use the discounted CKY tree aggregation approach instead of greedy top-down aggregation | Topic-segmentation-based tree generation · Preliminary experiments | Tree generation performance | Hurt |
Closest papers by meaning
- Hierarchical Macro Discourse Parsing Based on Topic Segmentation (AAAI 2021 · similarity 0.90)
- Unsupervised Learning of Discourse Structures using a Tree Autoencoder (AAAI 2021 · similarity 0.90)
- Top-Down RST Parsing Utilizing Granularity Levels in Documents (AAAI 2020 · similarity 0.88)
- A Deep Sequential Model for Discourse Parsing on Multi-Party Dialogues (AAAI 2019 · similarity 0.87)
- Narrowing the Gap between Supervised and Unsupervised Sentence Representation Learning with Large Language Model (AAAI 2024 · similarity 0.83)
- On the Affinity, Rationality, and Diversity of Hierarchical Topic Modeling (AAAI 2024 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.