Learning from Incomplete Ratings Using Non-negative Matrix Factorization

By Sheng Zhang, Weihong Wang, James C. Ford, Fillia Makedon. SDM, 2006

414 citations recorded by the source corpus.

Subfields: Artificial Intelligence, Information Systems

View Paper: https://www.semanticscholar.org/paper/f3a3deb3a93fa82ad647349e461deb7698f25932

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

1. Hamed Zamani
2. W. Bruce Croft
3. Jelena Kovacevic
4. José M. F. Moura
5. Siheng Chen
6. Aliaksei Sandryhaila
7. Azadeh Shakery
8. J. Dadashkarimi

In Information Systems

1. Hamed Zamani
2. W. Bruce Croft
3. Xing Xie
4. J. Dadashkarimi
5. Azadeh Shakery
6. Yongfeng Zhang
7. Fuzhen Zhuang
8. Hui Xiong

Citations in the corpus

• Pseudo-Relevance Feedback Based on Matrix Factorization (2016, background): “NMF has been shown to be highly effective in collaborative recommender systems [18, 44].”
• A Survey on Knowledge Graph-Based Recommender Systems (2020, methodology): “The item-item regularization is integrated with the weighted non-negative matrix factorization method [108] to refine low-rank representation of users and items for better recommendation.”
• Signal Recovery on Graphs: Variation Minimization (2014, methodology): “Those algorithms include SoftImpute [56], OptSpace [39], singular value thresholding (SVT) [57], weighted non-negative matrix factorization (NMF) [58] and graph-based weighted nonnegative matrix factorization (GWNMF) [59].”
• Temporal Pattern-Aware QoS Prediction via Biased Non-Negative Latent Factorization of Tensors (2020, background): “M1 (A Non-Negative Matrix Factorization Model [51], [52]): It considers a user–service–time tensor as a set of user– service matrix slices along the time dimension.”
• Nonnegative Matrix Factorization: A Comprehensive Review (2013, methodology): “An alternative suggested by Zhang et al. is to employ the EM algorithm where missing entries are replaced by the corresponding values in the current model estimate at the E-step and the standard unweighted multiplicative update rules are applied on the filled-in matrix at the M-step [109].”
• Federating Recommendations Using Differentially Private Prototypes (2020, background): “Such a constraint often results in more interpretable latent factors and improved performance (Zhang et al., 2006; McAuley and Leskovec, 2013).”
• An Efficient Non-Negative Matrix-Factorization-Based Approach to Collaborative Filtering for Recommender Systems (2014, methodology): “Suggested by Zhang et al. [27], this adaptation can be implemented through introducing either an EM procedure or indicator matrix into the NMF process.”
• Efficient retrieval of recommendations in a matrix factorization framework (2012, methodology): “For our situation, we can scale our dataset such that ∀ q ∈ S, ∥q∥ ≤ 1 and assume that the data is in the first quadrant (such as in non-negative matrix factorization models [19]).”
• Recommender Systems (2016, neutral): “The works in [252, 666] are non-negative forms of matrix factorization.”
• Explainable Recommendation via Multi-Task Learning in Opinionated Text Data (2018, methodology): “Various algorithms such as collaborative filtering [4, 18, 28, 29, 37] and factorization based methods [18, 22, 27] have been proposed to optimize those metrics.”