Personalization of Digital Content Recommendations

Personalization techniques for digital content recommendations are described. In one example, a hybrid model is used to form recommendations for individual users, groups of individual users, and so on. The hybrid model may also employ a latent factor model, which is configured to employ an implicit similarity approach to recommendations. The recommendations formed by these models are then used to generate a third, final, recommendation. As part of this, a weighting may be employed to weight a contribution of recommendations from the collaborative filter model and latent factor model in order to further personalize a recommendation for a user. Moreover, through application of localized regularization, for which every user is treated separately and also every content is considered independently, more personalization is achieved.