Front-end (2 UI's, ABBA testing):
- Statistics UI ( A): picasso.experiments.cs.kuleuven.be:3470?ui=statistics
- Sports coach UI ( B): picasso.experiments.cs.kuleuven.be:3470?ui=sportscoach
Back-end: https://sportify-hci.ew.r.appspot.com/. You can test an endpoint: https://sportify-hci.ew.r.appspot.com?most-popular-sports . Beware it can take a few seconds foor google cloud to spin up an instance.
The application consists of 2 parts:
- Front-end, deployed on picasso server.
- back-end api with noSQL firestore database, deployed on google cloud.
The front-end is a simple react app, deployed on the picasso server, which contains a Dockerfile and docker-compose that runs the entrypoint (clone git repo). The back-end is the api that does the recommendations, it uses a simple content-based recommender to calculate a cosine similarity factor between tags of the chosen 'preferences' sports and every other sport, and returns the top 5.
Both can be run locally if you wish, by navigating to the relevant subdirectory (front-end or back-end) and
executing npm start in both directories, then uncomment and change the back_end url in the auth.js so the front end will point to the local back end instead of the deployed one. Although both are deployed so this isn't necessary.
the 2 UI's are basically the same, the only difference is that the recommendations page renders a different component. When logging in, just specify a valid KU Leuven email adress:
Then, enter a random password. If the user doesn't exist, it is created and logged in.