Recommender System 101

How do most recommender systems suggest items to users?

• By showing random items

• By predicting scores and sorting items

• By using a fixed list of items

• By following manual input

Which metric can be used to measure a recommender system’s success?

• User Satisfaction

• Customer Loyalty

• Click Probability

• GMV

What is a common problem in recommender systems?

• Hallucination

• Out of Memory

• Confusion

• Data Sparsity

How does Collaborative Filtering (CF) make recommendations

• By comparing users with similar preferences

• By using pre-set rules

• By analyzing demographic features

• By trial and error

How is user-user similarity calculated in Collaborative Filtering?

• By comparing demographic data

• By hypothesis testing

• By optimizing parameters

• By using Pearson correlation of previous ratings

What is the time complexity of a basic Collaborative Filtering algorithm?

• N²M

• Nlog(M)

• NM

• NK + MK

How does Matrix Factorization (MF) differ from Collaborative Filtering?

• MF has more parameters

• MF uses SVD algorithm to train

• MF can generalize better

• MF is a memory based

Why is Matrix Factorization often used in state-of-the-art recommender systems?

• It converges instantly

• It is flexible

• It reduces data sparsity issues

• It is expensive to train