Papers › RST Parsing from Scratch
RST Parsing from Scratch
Thanh-Tung Nguyen, Xuan-Phi Nguyen, Shafiq Joty, XiaoLi Li
We introduce a novel top-down end-to-end formulation of document-level discourse parsing in the Rhetorical Structure Theory (RST) framework. In this formulation, we consider discourse parsing as a sequence of splitting decisions at token boundaries and use a seq2seq network to model the splitting decisions. Our framework facilitates discourse parsing from scratch without requiring discourse segmentation as a prerequisite; rather, it yields segmentation as part of the parsing process. Our unified parsing model adopts a beam search to decode the best tree structure by searching through a space of high-scoring trees. With extensive experiments on the standard English RST discourse treebank, we demonstrate that our parser outperforms existing methods by a good margin in both end-to-end parsing and parsing with gold segmentation. More importantly, it does so without using any handcrafted features, making it faster and easily adaptable to new languages and domains.
In Syntology Open this paper in Syntology's Atlas, the map of the papers in Syntology's graph and their citations.
Code
Repository list and official/mentioned flags are the archive's, frozen 2025-07-28. Reachability, where shown, is from one Syntology probe window (2026-09-16 to 2026-09-18); repositories not probed show nothing. GitHub stars are not tracked.
Code Syntology ran Syntology
Not run by Syntology. Nothing on this page verifies that the listed code works.
Tasks
Results from the paper archive 2025-07-28
| Task | Dataset | Model | Metric | Value | Rank at snapshot | Leaderboard | Report |
|---|---|---|---|---|---|---|---|
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | RST-Parseval (Nuclearity) | 76.0 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | RST-Parseval (Relation) | 61.8 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | RST-Parseval (Span) | 87.6 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | Standard Parseval (Full) | 50.2 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | Standard Parseval (Nuclearity) | 64.3 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | Standard Parseval (Relation) | 51.6 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (XLNet) | Standard Parseval (Span) | 74.3 | #17 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (Glove) | Standard Parseval (Full) | 46.8 | #21 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (Glove) | Standard Parseval (Nuclearity) | 59.6 | #21 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (Glove) | Standard Parseval (Relation) | 47.7 | #21 of 40 | Archive leaderboard | report |
| Discourse Parsing | RST-DT | End-to-end Top-down (Glove) | Standard Parseval (Span) | 71.1 | #21 of 40 | Archive leaderboard | report |
| End-to-End RST Parsing | RST-DT | Nguyen et al. (2021) | Standard Parseval (Full) | 46.6 | #4 of 4 | Archive leaderboard | report |
| End-to-End RST Parsing | RST-DT | Nguyen et al. (2021) | Standard Parseval (Nuclearity) | 59.1 | #4 of 4 | Archive leaderboard | report |
| End-to-End RST Parsing | RST-DT | Nguyen et al. (2021) | Standard Parseval (Relation) | 47.8 | #4 of 4 | Archive leaderboard | report |
| End-to-End RST Parsing | RST-DT | Nguyen et al. (2021) | Standard Parseval (Span) | 68.4 | #4 of 4 | Archive leaderboard | report |
Ranks are positions in the archive's leaderboards as they stood at the 2025-07-28 snapshot. Results published since then are not among these rows, so a rank here is not a current standing.
Methods
Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections