you could suggest music from artists that have released music on the same record label within +/-5 years of what you listened to and get close enough. the human curation is already "baked in".
graph traversal playlists are the most interesting idea to me, especially if you can put some bounds on (i.e. weight positively and negatively certain artists in the graph)
Graph visualizations are more useful for data science than end user tasks.
Even in lower cardinality contexts like music genres users are better served by traversing topics through a similarity mechanism (e.g. list of suggested other topics) as it gets to user benefit sooner. How often are users clicking far away topics from their area of interest? How do they know which direction to travel in?
Sufficiently large graphs are always subject to classification mechanisms of human benefit and cardinality. That means most graphs are best accessed through hierarchical interfaces.
One interesting thing is the weights of classification, e.g a band could play multiple genres but are they more representative of some than others. This might benefit from a creative interface to represent how dimensionally similar a topic is to some ideal entity in each classification value. Possibly a graph?
i'm going to create a startup for the ultimate integration between mobile phones and automobiles. it's this box and you put your phone in it, and while you're driving you can't open the box.
graph traversal playlists are the most interesting idea to me, especially if you can put some bounds on (i.e. weight positively and negatively certain artists in the graph)