The document proposes a novel latent factor model for recommender systems that is less complex than Funk SVD while maintaining comparable accuracy. It introduces the proposed model, which transforms users into a one-dimensional latent space and items into a multidimensional latent space, with ratings predicted as the product of the user's latent factor and the sum of the item's latent factors. An evaluation on movie and product rating datasets found the proposed model has comparable accuracy to Funk SVD but lower complexity due to requiring fewer latent features.