Item-MSMF: Items Most Similar based on Matrix Factorization
The Item-MSMF algorithm, this is recommender technique based on matrix factorization, that incorporates similarities of items which are calculated based on metadata. This approach to address the item cold-start through a shared latent factor vector representation of similar items based on those items which have enough interactions with users. In this way, the new items representations that are not accurate in terms of rating prediction, is replaced them with a weighted average of the latent factor vectors of the most similar items.
Item-MSMF: Items Most Similar based on Matrix Factorization
The Item-MSMF algorithm, this is recommender technique based on matrix factorization, that incorporates similarities of items which are calculated based on metadata. This approach to address the item cold-start through a shared latent factor vector representation of similar items based on those items which have enough interactions with users. In this way, the new items representations that are not accurate in terms of rating prediction, is replaced them with a weighted average of the latent factor vectors of the most similar items.