This repository accompanies the paper "A Multi-Domain System for Real-time Recommendations" submitted to ECIR'23 Demonstration Paper Track.

Demonstration goal: Setup an online movie recommender

The purpose of the following content is to show how a real-time movie recommender could be setup up and used in an online setting.

For this, we will use the MovieLens 100k dataset but the same workflow can be done for any other domain that contains item, user or interaction data.

In the Uptrendz platform, there are multiple ways on how to configure a multi-domain recommender system.
It is possible to create a domain on a system level (i.e., data between domains is separated and not shared) or on an item level (i.e., data is shared between domains). In this repository, we will create a movie recommender on the system level.

1. Go to uptrendz.ai
   • Register a new user account or log-in using your existing Google account
2. Create a new domain for MovieLens
   • We create a domain on the system-level
3. Configure the Item API
   • Need to define what meta-data is available for movies
4. Configure the User API
   • Need to define what meta-data is available for users
5. Configure the Interaction API
   • Need to define that we are using explicit rating interactions
6. Create recommendation scenarios
   • Need to create recommendation APIs with desired algorithms and post-processing rules
7. Run the provided notebooks to:
   • Upload the data (i.e., movies, users and ratings) to your newly created movie domain
     • Using the auto-generated data ingestion APIs
   • Request recommendations for the constructed recommendation scenario
     • Using the auto-generated recommendation APIs

Prerequisite:

In order to run the notebooks provided in this repository, you need to have a working jupyter notebook environment. In case you don't have one, the easiest way is to use pre-built docker images.

Example on Windows 11 with Ubuntu 20.04 using WSL2 and installed docker 20.10.18:

• Open Terminal
• If docker is not running, run sudo dockerd
• Run docker pull jupyter/scipy-notebook to retrieve the appropriate image
• Run docker run -it --rm -p 8888:8888 -v "$PWD"/notebooks/:/home/jovyan/ jupyter/scipy-notebook
• Open Jupyter lab by typing localhost:8888 in your browser (token password should be displayed in the terminal)