simple-recommender

A simple content based recommendation application. NLP model is trained in another repository. This application makes recommendation given input using pre-trained model.

Running application

Option 1: Docker

Requirements

Docker allows you to run this application on every operating system. The only requirement is to have running Docker engine instance. If it is installed and running, docker run hello-world command returns "Hello from Docker!" text. Otherwise, please head over to https://docs.docker.com/get-started/#download-and-install-docker and go through the installation steps.

Running

1. Download this repository to your local machine directly from here or via Git:

   git clone https://github.com/TUM-IDP-WS-20/simple-recommender.git

2. In the root directory, run docker-compose up

That's all you need to do! Docker build and run 3 containers:

• lrs_app : The main literature recommender system
• nginx : An Nginx Webserver to welcome requests first and redirect them to the LRS app running behind the nginx
• lrs_db : A Postgresql database to store user ratings

These all steps may take up to 5 minutes depending on your network connection. Once installation is done, you can reach the application on http://localhost.

Note: If you want to make changes on the code, this option has one drawback that you need to delete the image created for 'lrs_app' and rerun docker-compose up to build and run 'lrs_app' service.

Option 2: Setup Local Environment

1. Create a virtual environment

virtualenv --python=python3.8 .venv

2. Activate environment

source .venv/bin/activate

3. Install required python packages

pip3 install -r requirements.txt

4. Database Configuration

   1. Make sure that you have installed postgresql on your computer.
   2. Create a database with the following credentials:

      • Username: user

      • Password: password

      • Database Name: rec_database

      • Host: localhost

      • Port: 5432

      • Use docker/lrs_database_init.sql for initial tables

        OR simply run a prepared postgres container via:

        docker-compose run localDb bash

   3. Set environment variables

   export APP_SETTINGS="app.config.DevelopmentConfig"
   export DATABASE_URL="postgresql://user:password@localhost:5432/rec_database"

5. Running

   Run the command below to run the application:

   flask run

   You can reach the application on http://localhost:5000

Development

1. Install environment

2. Activate Github Large files to keep model files

   Install git lfs(Check up-to-date doc: https://git-lfs.github.com/)

   git lfs install
   git lfs pull

3. Database Configuration

   Make sure that you have installed postgresql.

   • docker-compose run localDb bash
   • export APP_SETTINGS="app.config.DevelopmentConfig"
   • export DATABASE_URL="postgresql://localhost:5432/rec_database"
   • python3 manage.py db init
   • python3 manage.py db migrate
   • python3 manage.py db upgrade

4. Run application

   Run the command below to run the application:

   flask run

   You can reach the application on http://localhost:5000

5. Keep dependencies clear

   All dependencies will be saved in requirements.txt. If a new library is needed for a notebook, please update dependecies on requirements.txt accordingly. You can follow the steps below to update it quickly:

   • Be sure to activate the environment as shown in 1.ii.
   • Install the packages that you need like pip3 install <package name>
   • Update the requirements.txt:

   pip3 freeze -r requirements.txt > requirements.txt

Deployment

1. Heroku Setup

   1. Access Create a Heroku account and install heroku CLI. You can follow instruction here: https://devcenter.heroku.com/articles/getting-started-with-python
   2. Login

      heroku login

   3. Add heroku remote upstream url

   git remote add heroku https://git.heroku.com/simple-recommender.git

   4. Fetch heroku branch

      git fetch heroku master

   5. Database Config
      1. heroku config:set APP_SETTINGS=app.config.ProductionConfig --remote heroku
      2. heroku addons:create heroku-postgresql:hobby-dev --app simple-recommender

2. Deployment

   1. Commit changes Merge changes to master branch
   2. Deploy Then, push changes to heroku to trigger new deployment:

      git push heroku master

3. Configure database if you made any change on it

   1. heroku config --app simple-recommender
   2. git push heroku master:master --no-verify
   3. heroku run python manage.py db upgrade --app simple-recommender