Factor in the neighbors: Scalable and accurate collaborative filtering

By Yehuda Koren. 2010

772 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems, Management Science and Operations Research

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Information Systems

In Management Science and Operations Research

Citations in the corpus

Why Amazon's Ratings Might Mislead You: The Story of Herding Effects (2014, neutral): “Indeed, models on collective intelligence, from majority voting to collaborating ﬁltering 9–12 and crowdsourcing, 13,14 all assume independent crowds, representing a critical gap between modeling frameworks and empirical insights.”
Learning Algorithms for Active Learning (2017, methodology): “For each movie, we pretrain an embedding vector by decomposing the full user/movie rating matrix using a latent factor model (Koren, 2010).”
Model recommendation for action recognition (2012, methodology): “We estimate these factors using Koren’s method [13]: 1.”
TrustSVD: Collaborative Filtering with Both the Explicit and Implicit Influence of User Trust and of Item Ratings (2015, methodology): “Specifically, Ma et al. (2008) propose a social regularization method (SoRec) by considering the constraint of social relationships. The idea is to share a common user-feature matrix factorized by ratings and by trust. Ma, King, and Lyu (2009) then propose a social trust ensemble method (RSTE) to li”
Dimensionality Reduction via Matrix Factorization for Predictive Modeling from Large, Sparse Behavioral Data (2015, background): “As explained by Koren (2010), there are two main approaches to collaborative ﬁltering problems: There are also a few papers where SVD was demonstrated to be inferior to non-DR techniques on particular data sets or settings, such as Huang et al. (2007) and Sun et al. (2005).”
Transfer Learning for Heterogeneous One-Class Collaborative Filtering (2016, background): “Intelligent recommendation technology [2] [3] [4] [5] [9] [10] [12] has been embedded in various online applications that connect people, products and services such as E-commerce, video streaming and social media sites.”
Explainable recommendation with fusion of aspect information (2018, methodology): “Collaborative Filtering (CF) based approaches [4, 15, 34, 38] make rating prediction using a decomposition of a user-item rating matrix, and fall into two categories, the memory based CF [6, 11, 12] and the model based CF [16, 17, 31].”
Semantic Matching in Search (2014, methodology): “Koren proposes factorized neighborhood model (FNM) for collaborative filtering, which enhances the traditional neighborhood model by factorizing the user-user and item-item similarities in the model [101, 102].”
On bootstrapping recommender systems (2010, methodology): “In particular, we concentrate on the factorized item-item approach [5], which combines close to the best known accuracy with signiﬁcant space and time-complexity improvements.”
Adaptive bootstrapping of recommender systems using decision trees (2011, methodology): “A common practice is assuming an additive parameterization of the bias, where bui = bu+bi [11, 15].”
Learning multiple-question decision trees for cold-start recommendation (2013, background): “…Decision tree Permission to make digital or hard copies of all or part of this work for personal or classroom use is granted without fee provided that copies are not made or distributed for profit or commercial advantage and that copies bear this notice and the full citation on the first page.”
entity2rec: Learning User-Item Relatedness from Knowledge Graphs for Top-N Item Recommendation (2017, methodology): “As baselines, we use state-of-the-art collaborative ltering algorithms based on Non-Negative Matrix Factorization (NMF) [17], on Singular Value Decomposition [15], ItemKNN with baselines [14] and the Most Popular Items recommendation strategy, which is known to be quite eective on MovieLens due to”
Factorization Machines with libFM (2012, methodology): “(47) |I||I| The FM model with this data would be equivalent to Factorized KNN mm 11 y (x)= y(i,{(r1,l1),...,(rm,lm)})= w0+ wi + rj wlj + rj(vi,vlj ) mm j=1 j=1 mm 1 + rjrjp(vlj ,vljp) (48) m2 j=1 jp>j This model is similar to the factorized nearest neighbor model [Koren 2010].”
Advances in Collaborative Filtering (2011, background): “Nonetheless, there is evidence that small improvements in RMSE terms can have a signiﬁcant impact on the quality of the top few presented recommendations [17, 19].”
Model recommendation: Generating object detectors from few samples (2015, methodology): “We use collaborative filtering techniques based on matrix factorization, which assume a low-rank approximation to the ratings store that naturally embeds both tasks and models to a joint latent factor space of dimensionality d, such that task-model interactions are modeled as inner products in that ”
Neural Networks and Deep Learning (2018, neutral): “An item-based autoencoder is discussed in [436], and this approach is a neural generalization of item-based neighborhood regression [253].”
Exploring Artist Gender Bias in Music Recommendation (2020, methodology): “get user u for a target artist a. We then evaluate RS recommending the top-5 artists with the highest predicted preferences. We consider two types of CF algorithms: (1) KNN-based approach: UserKNNAvg [27], and (2) factorisation-based approach: Non-Negative Matrix Factorization (NMF) [30]. Hyperparam”
Functional matrix factorizations for cold-start recommendation (2011, methodology): “Recently, matrix factorization becomes a popular direction for collaborative ﬁltering [12,14,16,21,24].”