If you want to use dataset of this project just enter:

$ python3 ./src/preprocess_data.py

It takes a while because scrapping is in process.

If you want to use your own dataset create urls.csv file inside data folder with the following data structure:

Values in url column are not required, but the column itself should be the 1st.

There should be at least 2 context columns. 1st context column must be the 4th in csv.

To install all dependencies (using venv) and train models do the following:

$ ./pipeline.sh

To predict category and context enter:

$ python3 ./src/predictor.py --url URL [--user USER]

URL must start with http(s)://

For example:

$ python3 ./src/predictor.py --url https://engadget.com --user office

The commard prints predicted values and True if office is in context else False.