The goal of this project is to improve the IBM Watson articles recommendation. At this project you will find a notebook showing the implementation of the 3 types of recommendation approaches most used, also there is a python module, where you can find all the functions used at the notebook encapsulate and ready for you to use.
This project is part of the Udacity Data Science Nanodegree program.
Create a virtual environment named ibm_venv.
$ python3 -m venv ibm_venv -- for Linux and macOS
$ python -m venv ibm_venv -- for Windows
After that, activate the python virtual environment
$ source ibm_venv/bin/activate -- for Linux and macOS
$ ibm_venv\Scripts\activate -- for Windows
Install the requirements
$ pip install -r requirements.txt
The two csvs needed to run the notebook are at the data folder, but because the process to make a recommendation using NLP is expensive, the user could download the resulting csv at the link below and put the file in the data folder.
https://drive.google.com/file/d/1jNjllAzcnM50nTHFJMx7Fv0QPVwm5d0R/view?usp=sharing
This csv is needed also to run the tests, by the way to run the tests you can execute the command:
python tests.py
Finally you can create a file script and import the Recommender from the recommender_template.py.
- The
data
folder contains the disaster's data and the script to clean and store the data. - The
requirements.txt
has the needed packages to run the code. Recommendations_with_IBM
notebook with all the processrecommender_functions
python script with all the functionsrecommender_template
python script with the Recommender classproject_tests
the project tests
Part of the code used is inspired from the Experimental Design & Recommendations module of the Udacity Nanodegree. Go ahead and contribute to this repository. the data was kindly provided by IBM Watson Studio Platform