SENTS is a Neural Text Simplification system using Semantic Splitting for: Simple and Effective Text Simplification Using Semantic and Neural Methods

• Python 3.6

1. Checkout the repository

git clone https://github.com/nzaim001/SENTS-reproduction.git
cd SENTS-reproduction/

2. Create a Python virtual environment. For example, on Linux:

virtualenv --python=/usr/bin/python3 venv
venv/bin/activate

3. Install tupa & matplotlib

pip install tupa matplotlib

4. Download and unpack TUPA's pre-trained models

curl -LO https://github.com/huji-nlp/tupa/releases/download/v1.3.3/ucca-bilstm-1.3.3.tar.gz
tar xvzf ucca-bilstm-1.3.3.tar.gz

5. OpenNMT dependencies

   1. Install Torch (Using conda: lua-torch)
   2. Install additional packages:

   luarocks install tds

6. Checkout the NTS model repository & Download the pretrained models NTS & NTS-w2v

git clone --recursive https://github.com/senisioi/NeuralTextSimplification.git
cd NeuralTextSimplification
python src/download_models.py ./models
cd ..

7. Apply the existing patch to the NTS model

patch -s -p0 < NTS.patch

8. Get help on using the Makefile to run the Text Simplification system

make help

1- Install the environment : Nodejs, AngularJS, necessary Modules/Packages:

cd website/
. install_modules.sh

2- Run the server :

node server.js

Then navigate to :

localhost:3000/

You can edit the web listening port in serve.js. In our case, it is 3000

• article_to_sentences.py: A small script that splits the given article to many files, each one containing a single sentence.
• split_sentences.py: A script that uses two semantic rules to split sentences presented in a UCCA-parsed XML format.

• A web server to run the Text Simplification system

• Hamza Ezzarouali El Boudri: [email protected]
• Nabil Zaim: [email protected]

• M. Olivier Augereau
• M. Kise Koichi
• IMP
• Osaka Prefecture University
• ENSEIRB-MATMECA | Bordeaux INP