wikipedia-article-summarization

Summarizes Wikipedia articles with transformers T5 model. You can execute it through a Python script, access it through a fast API's REST API or through a streamlit web page.
You can access a live web page here.

Installing

1. Install requirements

pip install -r requirements.txt

2 (optional). Install Docker if you wish to run the web application in a Docker container. You can skip step 1 if that is the only application you wish to execute.

Usage

1. Python program

Run summarize_wikipedia.py [--limit <character_limit>] <english wikipedia page's URL>:

python summarize_wikipedia.py --limit 1000 https://en.wikipedia.org/wiki/PageRank

2. REST API

Generate an API key (e.g. "abc" with salt "1"):

python ./utils/hash_key.py "abc" "1"

Set API key and salt as environment variables (use export instead of set for Linux systems):

set SUMMARIZE_API_KEYS="abc"
set SUMMARIZE_SALT="1"

Run fast API with uvicorn (or other web server of your choice):

uvicorn api:app --host 0.0.0.0 --port 8000

Send a GET request to /summarize/ endpoint with URL and character limit as parameters, with API key as header (/utils/request.ps1 provides a Powershell example):

./utils/request.ps1

3. Web app

You can run the web app through streamlit directly or through Docker using the provided Dockerfile.

Run streamlit web app (python):

streamlit run webpage.py

Run streamlit web app (docker):

docker build -t summarizer_webapp .

docker run -p 8502:8502 summarizer_webapp

Access the web page in http://localhost:8501 (ran via streamlit) or http://localhost:8502 (ran via docker).