Building the interface itself is the trivial part of this exercise. How are you going to source the database of metadata of every media item produced in the last century?
There are a several longstanding open projects out there to compile metadata databases that are free to use like: https://musicbrainz.org/statistics (33 million tracks)
Music is less than 100 million items, Movies fewer than 1 million, TV somewhere in between and books are a few hundred million. I'm sure they all have mostly-complete metadata databases available already.
So the answer is basically “just empower the community to do it”.