User modeling

Single prototype
Matrix factorization

Distributed representation
Variational auto-encoders:
 * "Variational autoencoders for collaborative filtering": Liang et al. (2018)
 * "queriable" variant that takes into account the ambiguity of people who have many interests (but still use a single Gaussian distribution to model them): Wu et al. (2019)

Multiple prototypes/clusters
Wu et al. (2019) demonstrate that users who interact with more items often have diverse interests and modeling these interests properly help improve the relevance of recommendations.

Content-based
From Wang et al. (2006) : Chen et al. (2001) and Pretschener et al. (1999) propose "a profile model with hierarchical concept categories"

Discrete presentation
From Si and Jin (2003) : probabilistic latent space model, which models individual preferences as a convex combination of preference factors. [...] The aspect model assumes that users and items are independent from each other given the latent class variable." collaborative filtering in (Hofmann & Puzicha, 1999) . This model assumes that each user should belong to exactly one group of users and the same is true for each item" Si and Jin (2003) : "FMM extends existing partitioning/clustering algorithms for collaborative filtering by clustering both users and items together simultaneously without assuming that each user and item should only belong to a single cluster"
 * "The aspect model (Hofmann & Puzicha, 1999) is a
 * "two-sided clustering model is proposed for
 * "In the personality diagnosis model (Pennock et al., 2000), the observed rating for the test user yt on an item x is assumed to be drawn from an independent normal distribution with the mean as the true rating"