Practise vocabulary for the cold start problem in recommendation systems: user cold start, item cold start, onboarding strategies, fallback recommendations, and popularity bias.
0 / 5 completed
1 / 5
The 'cold start' problem in recommendation systems refers to difficulty recommending for ___ with little or no interaction history.
Cold start occurs when a recommendation system lacks the interaction data needed to personalise — either because the user is new (no past behaviour) or the item is new (no ratings/interactions from other users). Collaborative filtering is most affected because it relies entirely on historical interaction data.
2 / 5
A ___ cold start occurs when a brand-new user signs up and the system has no interaction history to base recommendations on.
User cold start means no purchase history, no ratings, no clicks for this person. Common mitigations: onboarding surveys ('pick 5 topics you like'), demographic-based recommendations, or defaulting to popularity-based fallbacks until enough interactions accumulate (typically 5–20 actions).
3 / 5
A ___ cold start occurs when a newly added product or article has received no ratings or interactions yet.
Item cold start means collaborative filtering cannot include the new item because no user has interacted with it — it won't appear in any similarity calculation. Mitigations: content-based filtering (use item metadata like category, tags), editorial boosting (manually surfacing new items), or exploration strategies that randomly expose new items to a fraction of users.
4 / 5
A 'new user onboarding strategy' in recommendation systems typically involves ___.
Onboarding surveys ('what are you interested in?') collect explicit preference signals to bootstrap the user profile. Even 3–5 explicit signals allow content-based filtering to start making relevant recommendations, avoiding the awkward 'no recommendations available' state and improving early retention.
5 / 5
'Popularity bias' in recommendation systems means the algorithm ___ popular items at the expense of niche or new items.
Popularity bias is a self-reinforcing loop: popular items get more interactions → more interactions make them score higher → they get recommended more → they get more interactions. This makes it harder for new or niche items to gain exposure. The cold start problem and popularity bias are connected: cold-start fallbacks (show popular items) amplify popularity bias. Mitigation strategies include diversity injection, inverse popularity weighting, and exploration bonuses for new items.