Docker image: ghcr.io/rst-workbench/rst-converter-service
REST API to convert between different Rhetorical Structure Theory file formats. It is built on top of the discoursegraphs library.
- CODRA
- dis (e.g. used by RST-DT corpus)
- DPLP
- Heilman/Sagae (2015)
- rs3
- HILDA (also used by Feng/Hirst 2014)
- Note: We don't use the original HILDA format, but a slightly adapted one based on the internal nltk.Tree representation.
- arne-cl/feng-hirst-rst-parser, nlpbox/feng-hirst-service and nlpbox/hilda-service can produce this format.
(Contrast[S][N]
_!Although they did n't like it ,!_
_!they accepted the offer . <P>!_)
ParseTree('Contrast[S][N]', ["Although they did n't like it ,", 'they accepted the offer .'])
- dis
- rs3
- rstlatex (for embedding RST trees into LaTeX documents)
- tree.prettyprint (ASCII-style tree)
- svg (SVG image of an nltk Tree)
- svg-base64 (base64 encoded SVG image of an nltk Tree)
Note: The old output format names svgtree and svgtree-base64 are deprecated
aliases of svg and svg-base64. They still work, but responses are flagged with a
Deprecation header and the aliases will be removed in a future release.
The simplest way to install the rst-converter-service is using Docker:
docker pull ghcr.io/rst-workbench/rst-converter-service:latest
Alternatively, build the image from source:
git clone https://github.com/rst-workbench/rst-converter-service.git
cd rst-converter-service/
docker build -t rst-converter-service .
To run the web service, type:
docker run -p 5000:5000 -ti ghcr.io/rst-workbench/rst-converter-service:latest
In another terminal, you can now convert RST files. To convert the file car-repair.rs3
from rs3 format to dis format, type:
curl -XPOST localhost:5000/convert/rs3/dis -F input=@car-repair.rs3
(Root
(span 1 2)
(Satellite
(leaf 1)
(rel2par background)
(text
_!I am having my car repaired in Santa Monica (1522 Lincoln Blvd.) this Thursday 19th._!))
(Nucleus
(leaf 2)
(rel2par span)
(text
_!Would anyone be able to bring me to ISI from there by 5 pm please?_!)))
For "visualizing" the RST tree, you might try prettyprinted trees:
curl -XPOST localhost:5000/convert/rs3/tree.prettyprint -F input=@car-repair.rs3
background
_____________|______________
S N
| |
I am having my Would anyone be
car repaired in able to bring me
Santa Monica ( to ISI from
1522 Lincoln there by 5 pm
Blvd.) this please?
Thursday 19th.
For a graphical rendering of the RST tree, use the svg output format:
curl -XPOST localhost:5000/convert/rs3/svg -F input=@car-repair.rs3 -o car-repair.svg
Here's what the conversion of short.rs3 from tests/fixtures/input looks like:
To see all supported input and output formats, type
curl localhost:5000/input-formats
["codra", "dis", "dplp", "hilda", "hs2015", "rs3"]
or
curl localhost:5000/output-formats
["dis", "rs3", "rstlatex", "svg", "svg-base64", "tree.prettyprint"]
If you use the rst-converter-service in your academic work, please cite the following paper:
Neumann, A. 2015. discoursegraphs: A graph-based merging tool and converter for multilayer annotated corpora. In Proceedings of the 20th Nordic Conference of Computational Linguistics (NODALIDA 2015), pp. 309-312.
@inproceedings{neumann2015discoursegraphs,
title={discoursegraphs: A graph-based merging tool and converter for multilayer annotated corpora},
author={Neumann, Arne},
booktitle={Proceedings of the 20th Nordic Conference of Computational Linguistics (NODALIDA 2015)},
pages={309-312},
year={2015}
}