Can Latent Features Be Interpreted as Users in Matrix Factorization-Based Recommender Systems?
Armelle BrunMarharyta AleksandrovaAnne BoyerPublished in: WI-IAT (1) (2014)
Keyphrases
- recommender systems
- matrix factorization
- collaborative filtering
- latent factors
- user ratings
- cold start problem
- data sparsity
- implicit feedback
- user preferences
- low rank
- recommendation quality
- user profiles
- information overload
- factorization methods
- rating prediction
- item recommendation
- cold start
- user model
- recommendation systems
- latent factor models
- variational bayesian
- factor analysis
- negative matrix factorization
- nonnegative matrix factorization
- stochastic gradient descent
- feature extraction
- feature space
- recommendation algorithms
- user interests
- feature vectors
- user feedback
- singular value decomposition
- missing data
- image features
- personalized recommendation
- user behavior
- co occurrence
- web search
- probabilistic matrix factorization