https://github.com/bact/incidentamr-server
A server for IncidentAMR
https://github.com/bact/incidentamr-server
abstract-meaning-representation incident-reports natural-language-processing
Last synced: 3 months ago
JSON representation
A server for IncidentAMR
- Host: GitHub
- URL: https://github.com/bact/incidentamr-server
- Owner: bact
- License: mit
- Created: 2023-12-01T04:09:55.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-12-15T09:30:38.000Z (over 1 year ago)
- Last Synced: 2025-02-24T12:46:39.372Z (3 months ago)
- Topics: abstract-meaning-representation, incident-reports, natural-language-processing
- Language: HTML
- Homepage:
- Size: 72.3 KB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# incidentamr-server
A server to test parsing text to Abstract Meaning Representation. All of these are for experimentation purpose. I would like to develop a small meaning representation corpus for incident report in the future.
- Use [AMRLib](https://github.com/bjascob/amrlib) for AMR parsing, with a Sentence to Graph model from [amrlib-models](https://github.com/bjascob/amrlib-models) (AMR is a graph, rather than a tree)
- Use [Uvicorn](https://www.uvicorn.org/) and [FastAPI](https://fastapi.tiangolo.com/) for web server and API framework
## Install
1. Download source
```sh
git clone https://github.com/bact/incidentamr-server.git
```
2. Install required libraries
```sh
cd incidentamr-server
pip install -r requirements.txt
```
3. Install a model
- Download any *Sentence to Graph* model from [amrlib-models/releases](https://github.com/bjascob/amrlib-models/releases).
- For example, `model_parse_xfm_bart_large-v0_1_0.tar.gz`.
- Extract the tar.gz file, you will get a directory containing .json and .bin files. Rename that directory to `stog`.
- Put the `stog` directory inside `incidentamr-server/incidentamr_server/models` directory.
## Run
1. Start the server. From inside `incidentamr_server` directory, run:
```sh
uvicorn main:app --reload
```
2. Then, from within a web browser, open [http://127.0.0.1:8000](http://127.0.0.1:8000).
The web interface will look like this:
## Meaning Representations
The main paper for Abstract Meaning Representatio (AMR) is
- Banarescu, Laura, Claire Bonial, Shu Cai, Madalina Georgescu, Kira Griffitt, Ulf Hermjakob, Kevin Knight, Philipp Koehn, Martha Palmer, and Nathan Schneider. 2013. **‘Abstract Meaning Representation for Sembanking’**. In Proceedings of the 7th Linguistic Annotation Workshop and Interoperability with Discourse, 9. Sofia, Bulgaria: Association for Computational Linguistics. https://aclanthology.org/W13-2322.pdf.
More information about AMR can be found at [AMR Bank](https://amr.isi.edu/) website.
Another interesting development is Uniform Meaning Representation (UMR). Their main paper is
- Van Gysel, Jens E. L., Meagan Vigus, Jayeol Chun, Kenneth Lai, Sarah Moeller, Jiarui Yao, Tim O’Gorman, et al. 2021. **‘Designing a Uniform Meaning Representation for Natural Language Processing’**. KI - Künstliche Intelligenz 35 (3): 343–60. https://doi.org/10.1007/s13218-021-00722-w.
More information about UMR can be found at [UMR Project](https://umr4nlp.github.io/web/).
*U in UMR in this meaning representation area can be many things: Unified, Uniform, Universal, etc.