• Track is dependent on Album. Can't exist without it (in this data model).

  • Adopting album PK as part of Track PK reflects this.

  • Could have invented track id as column, made it PK, but still need the album_id foreign key.

  • Album is dependent on Artist, so it should be possible to get rid of album_id. I.e., instead of identifying Album by album_id use (artist_id, title). (Assuming artists don't ever reuse album titles.)