Ecosyste.ms: Awesome
An open API service indexing awesome lists of open source software.
https://github.com/nusnlp/esr
Last synced: 17 days ago
JSON representation
- Host: GitHub
- URL: https://github.com/nusnlp/esr
- Owner: nusnlp
- License: gpl-3.0
- Created: 2021-09-09T04:47:01.000Z (over 3 years ago)
- Default Branch: main
- Last Pushed: 2022-02-24T12:13:56.000Z (almost 3 years ago)
- Last Synced: 2023-08-21T05:20:51.583Z (over 1 year ago)
- Language: Python
- Size: 35.2 KB
- Stars: 11
- Watchers: 2
- Forks: 2
- Open Issues: 1
-
Metadata Files:
- Readme: README.md
- License: LICENSE
Awesome Lists containing this project
README
# Improved Word Sense Disambiguation with Enhanced Sense Representations
[](https://paperswithcode.com/sota/word-sense-disambiguation-on-supervised?p=improved-word-sense-disambiguation-with)
This repository contains codes and scripts to build enhanced sense representations for word sense disambiguation.
If you use this code for your work, please cite this [paper](https://aclanthology.org/2021.findings-emnlp.365.pdf):
```
@inproceedings{song-etal-2021-improved-word,
title = "Improved Word Sense Disambiguation with Enhanced Sense Representations",
author = "Song, Yang and
Ong, Xin Cai and
Ng, Hwee Tou and
Lin, Qian",
booktitle = "Findings of the Association for Computational Linguistics: EMNLP 2021",
year = "2021",
url = "https://aclanthology.org/2021.findings-emnlp.365",
pages = "4311--4320"
}
```
Requirements
------------
* python==3.8.8
* pytorch==1.9.0
* transformers==4.6.1
* nltk==3.6.2
Downloading Datasets
------------
You need to download the following datasets:
* [WSD Evaluation Framework](http://lcl.uniroma1.it/wsdeval)
* [UFSAC](https://drive.google.com/file/d/1Oigo3kzRosz2VjyA44vpJZ58tDFyLRMO)
* [FEWS](https://nlp.cs.washington.edu/fews/)
Setting up variables
------------
You need to modify `script/config.sh` according to your environment.
Set `data` variable to the top directory where all the datasets are stored.
Processing FEWS
------------
```
bash experiment/fews/run.sh
```
Using trained models
------------
You can train the models from scratch.
Alternatively, you can use our [trained models](https://drive.google.com/file/d/1c8yooOoXsnIgJi0-To7xKNmYU-CugaeL/view?usp=sharing).
Running Experiments
------------
For ESR on SemCor with `roberta-base`:
```
bash experiment/esr/roberta-base/dataset_semcor/sd_42/run.sh
```
For ESR on SemCor with `roberta-large`:
```
bash experiment/esr/roberta-large/dataset_semcor/sd_42/run.sh
```
For ESR on SemCor and WNGC with `roberta-base`:
```
bash experiment/esr/roberta-base/dataset_semcor_wngc/sd_42/run.sh
```
For ESR on SemCor and WNGC with `roberta-large`:
```
bash experiment/esr/roberta-large/dataset_semcor_wngc/sd_42/run.sh
```
For ESR on FEWS with `roberta-base`:
```
bash experiment/esr/roberta-base/dataset_fews/sd_42/run.sh
```
For ESR on FEWS with `roberta-large`:
```
bash experiment/esr/roberta-large/dataset_fews/sd_42/run.sh
```