Recommendation engines
Recommendation engines usually rely on either attributes of the thing itself (feature-based) or how people interact with the thing (collaborative filtering).
Example:
Feature-based: You enjoyed this country song — here are more country songs you might like.
Collaborative filtering: You enjoyed song X — here are more songs that listeners of song X also enjoyed.
The feature-based approach anchors the recommendation based on the genre, while the collaborative filtering approach bases the rec off user behavior.
When it comes to finding new movies, I’ve found success with looking up other movies that that share the same director or producer. This is effectively a manual feature-based recommendation.
This feels like a more human way to find new content. And if the director has a long career, dabbling across multiple genres or time periods, it’s a good way to escape the filter bubble that recommendation engines are known for creating.
Because this approach is manual and not automated, my ideal interface for this approach is a network diagram — kind of like a mind map.
In the above network diagram, we can see the connection between 2 of my favorite movies Get Out and Donnie Darko.