vTLDR

Using PhoBERT pretrain, vTLDR provides a seq2seq model that performs the mini downstream task of Vietnamese news summarization. A GUI for users to input plain text or news URL (only supports tuoitre.vn domain at the moment) and output respective summary.

Notebook train_test_infer guide: Colab App demo video: Youtube

Usage

Installation

Clone repo

git clone https://github.com/ngfuong/vTLDR
cd vTLDR

Create virual environment and install requirements

python -m venv ./venv
source ./venv/bin/activate
pip install -r requirements.txt

App

Download checkpoint from Drive and put into training folder.

Make sure vncorenlp folder have the .jar file and its word segmentation component. You can download them as follow

mkdir -p vncorenlp/models/wordsegmenter
wget https://raw.githubusercontent.com/vncorenlp/VnCoreNLP/master/VnCoreNLP-1.1.1.jar
wget https://raw.githubusercontent.com/vncorenlp/VnCoreNLP/master/models/wordsegmenter/vi-vocab
wget https://raw.githubusercontent.com/vncorenlp/VnCoreNLP/master/models/wordsegmenter/wordsegmenter.rdr
mv VnCoreNLP-1.1.1.jar vncorenlp/ 
mv vi-vocab vncorenlp/models/wordsegmenter/
mv wordsegmenter.rdr vncorenlp/models/wordsegmenter/

To use app, launch

python app.py

Plain text or news URL (tuoitre.vn domain) are supported as input.

Training and Testing

For training, download Vietnews dataset

wget https://github.com/ThanhChinhBK/vietnews/archive/master.zip
unzip master.zip

Make sure the dataset is processed, refer to data_preprocess.py. Train with default config.yaml file. Edit the file to change training configs.

python train.py

To test

python test.py

References

1. Nguyen, Van-Hau & Nguyen, Thanh-Chinh & Nguyen, Minh-Tien & Hoai, Nguyen. (2019). VNDS: A Vietnamese Dataset for Summarization. 375-380. 10.1109/NICS48868.2019.9023886.
2. Rothe, Sascha & Narayan, Shashi & Severyn, Aliaksei. (2020). Leveraging Pre-trained Checkpoints for Sequence Generation Tasks. Transactions of the Association for Computational Linguistics. 8. 264-280. 10.1162/tacl_a_00313.
3. Nguyen, Dat Quoc & Nguyen, Anh. (2020). PhoBERT: Pre-trained language models for Vietnamese. 1037-1042. 10.18653/v1/2020.findings-emnlp.92.
4. ngockhanh5110, nlp-vietnamese-text-summarization, (2021), GitHub repository, https://github.com/ngockhanh5110/nlp-vietnamese-text-summarization