New Muses Project makes composer recommendations! Here's how it works.
When you give us the name of a composer you like, a chain-reaction of tasks are set off.
First, we dig into our database to locate the composer you're thinking of. Say you've typed in Johann Sebastian Bach.
Then, we see which other composers Bach is connected to. We've curated each of these connections based on parameters such as historical era, influences, genres composed, location, style, and sound. We focus on composers that have been systematically excluded from the canon.
Finally, we pick one of these connections at random to suggest back to you!
The nitty gritty stuff (for technically-minded and curious readers!)
Our database is structured as a directed graph, with vertices (nodes) representing composers and directed edges representing recommendations. The composers we want to recommend are labelled "interesting" and colored purple. All composers in our database may have recommendations, but only "interesting" composers may receive recommendations.
Since composers will typically be connected with others who lived at a similar time, our graph lends itself to being organized chronologically. Visualizing music history in this way encompasses a complexity beyond the traditional lineage of classical composers. We think it's beautiful.
Zooming in, we can also visually locate clusters of composers who might be considered "similar," or who at least may have known each other in life. Eventually we'll hope to run a density-based or pattern-based clustering algorithm to programatically detect these communities!