FiML
Link to github project page: here
This is a web app that leverages the recommender models trained using RecLibWH, to recommend movies. I wanted this app to set itself apart from the available recommendation platforms out there, by fulflling the following design objectives:
- Being highly customizable. The app should be able to respond to a user’s instantaneous preferences, instead of returning a fixed list of mixed recommendations every time.
- Transparent. Allowing users to choose exactly what information goes into each recommendation. This point also plays into customizability as it allows users to tune inputs into the recommender.
- Highly responsive. Every rating / interation is taken into account instantaneously, in real time.
With the above objectives in mind, I created FiML - the Machine Learning Film recommender app. The frontend and backends are powered by React and Django respectively, and leverages an adaptive variant of the 2007 Netflix Prize-winning recommender model.
The location of the live version is available upon request, since at this time, my resources for hosting the machine learning model is quite limited, and I would like to control the amount of users on the platform at any one time. I am actively working on trying to scale up the deployed version, however, so please stay tuned.