AI in Music Recommendation Systems Quiz

Answers and explanations appear as you take the quiz.

1. What is the 'cold start' problem in music recommendation?
   • Difficulty recommending brand-new tracks or users with no listening history
   • Songs that play too slowly
   • Servers crashing in winter
   • Playlists that are too long
2. How can content-based audio analysis help recommend a new or little-heard song?
   • By relying only on the new track’s total number of past plays
   • By comparing audio-derived representations with tracks that match a listener’s musical preferences
   • By matching its file size to songs in the same playlist
   • By using the song’s upload date as its only similarity signal
3. Collaborative filtering primarily works by finding what?
   • Songs of the same length
   • The cheapest tracks
   • Listeners with similar habits and recommending what they liked
   • Songs in alphabetical order
4. In embedding-based systems, two songs are considered similar when their vectors are what?
   • The same length string
   • Different colors
   • Stored on the same server
   • Close together in the shared space
5. What does Pandora's Music Genome Project rely on heavily?
   • Detailed musical attribute tagging of songs
   • Random shuffling
   • Only the artist's popularity
   • The user's email address
6. Why are platforms paying more attention to fairness in recommendations?
   • To make playlists shorter
   • To avoid only reinforcing a few mega-hits and help smaller artists get discovered
   • To reduce file sizes
   • To slow down playback
7. What does playlist co-occurrence contribute to a music recommender?
   • A measurement of the track’s audio sample rate
   • A prediction of how much storage a song requires
   • Context about which tracks listeners place together in playlists
   • A replacement for listening-history data
8. Why is nearest-neighbor search useful after music tracks are mapped into an embedding space?
   • It transcribes every track’s lyrics first
   • It changes the recording’s sample rate
   • It selects songs by file name only
   • It can quickly retrieve tracks whose vectors are close to a user’s vector