covid-19-applet

Installation

Create folder

mkdir c19_project
cd c19_project

Clone applet repository

git clone https://github.com/MrMimic/covid-19-applet

Check Python'v version

python3 --version

Should be >= 3.6.1. Update if needed.

Create venv

python3 -m venv venv
source venv/bin/activate

Download Kaggle dataset

Use the kaggle API:

pip install kaggle
kaggle datasets download allen-institute-for-ai/CORD-19-research-challenge

Create the local database

Clone the lib repository:

git clone https://github.com/MrMimic/covid-19-kaggle

Install the developped kaggle lib:

pip install -q git+https://github.com/MrMimic/covid-19-kaggle

Make modifications in the script: covid-19-kaggle/src/main/scripts/create_db.py

• L:17 local_path="articles_database.sqlite"
• L:18 kaggle_data_path="kaggle_data"

Should be changed by your local paths to wanted DB file and downloaded dataset.

At the moment, let run_on_kaggle and only_covid params to True, otherwise the resulting DB will weight 22Go instead of 700Mo.

Run the database creation:

python3 covid-19-kaggle/src/main/scripts/create_db.py

Install web applet

pip install poetry
poetry install

Run applet

Make change to server.py. It needs the DB you just created and the path to the pre-trained word2vec coming with the repository of the lib code:

• L15: LOCAL_DB_PATH = "<path_to_trained_db>"
• L16: LOCAL_EMBEDDING_PATH = "<path_to_cloned_library>/covid-19-kaggle/resources/global_df_w2v_tfidf.parquet"

Then run server:

python3 server.py

Go to http://:5000.

Watch out, data are stored in cache for only 10min right now. First time you launch the server and a new query after 10 min of inactivity will last ~30sec.

Troubleshot

• Open port if distant host