Movie Recommender

This repository contains the code implementation and demonstration of our movie recommending service. Two algorithms,
corresponding to two different approaches have been implemented:

1. Hybrid
2. Neural Collaborative Filtering (NCF)

Installation & Setup

Environment

We are using Python 3 (more specifically, Python 3.8). Nevertheless, any Python>=3.5 should work fine since we are not
using any version-specific functionalities.

Install the dependencies

pip install matplotlib numpy pandas scikit-learn scikit-surprise Flask flask-restful flask-cors torch torchvision pytorch-lightning

The above command will install the most recent version of the Python dependencies. If that creates any problems, we have
provided a requirements.txt file so that the exact environment can be emulated.

Download data and model files

1. Download the data files - movies.csv, ratings.csv, tags.csv - from the link below, and place it in the project root
   folder:
   Data files
   You can also download these files directly from the GroupLens official website:
   GroupLens MovieLens Dataset
   Just unzip the files and place them in the project root folder.

2. (Optional) If you do not want to train the model yourself, we have provided pre-trained model so that you can quickly
   get started. Place the trained.model in the code/ncf folder, and the hybrid.model in the code/hybrid
   folder.
   trained.model (size ~800 MB)
   hybrid.model (size ~130 MB)

Train the models

1. To train the NCF model simply run the code/ncf/Train.py file. This will take some time. The model that will be
   generate will be placed in the same folder.

2. To train the Hybrid model, run the code/hybrid/train.py file. The trained model will be generated and placed in the
   same folder.

Run the application

To run the flask application, execute the main method present in code/ncf/main.py.

This should start the flask application on Port:5000 of your system.

Caveats

1. We observed that PyTorch installation using pip sometimes causes problems. In that case, follow the instructions on
   the PyTorch Official Website to install the version that is right for you.

2. PyTorch installation using pip sometimes gives some setup.py errors. Just verify that your Python installation is
   a 64-bit installation. PyTorch does not work on a 32-bit Python installation on Windows.

UI Application Installation and Setup

Install the dependencies

1. Install Node.js from: https://nodejs.org/en/download/
2. Open terminal in code/UI folder and run the command npm install to install all the dependent packages.

Run the app

1. Open terminal in the code/UI folder and run the command ng serve, this will run the application and serve it on Port : 4200
2. Open your browser and navigate to http://localhost:4200

User Id's for testing

When you open the UI application, you will be required to enter the User Id for the user you want to get the recommendations for.

If you use the pre-trained model files, you could use the following user id's to get both NCF and Hybrid recommendations: 2103, 5413, 12, 18768.