Pith. sign in

REVIEW 3 major objections 4 minor 70 references

Towards Reliable Recommender Systems for Rating Data

T0 review · 3 major / 4 minor · reviewed 2026-08-10 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read This paper shows that robust discrete matrix completion (RDMC) keeps recommender predictions stable under fake-profile attacks, while sacrificing little accuracy in clean data.

desk verdict Useful robust-discrete matrix completion paper whose headline robustness claim is confounded by median versus mean centering and needs an ablation before the mechanism is clear. read the letter →

arxiv 2412.20802 v3 pith:HRJECL4O submitted 2024-12-30 stat.ML cs.LGstat.COstat.ME

Aurore Archimbaud , Andreas Alfons , Ines Wilms This is my paper · ORCID
classification stat.MLcs.LGstat.COstat.ME MSC 62F3562H1268T05
keywords robustmatrixcompletiondiscreterating-scaledatarecommendersystemsadversarialattacksfakeprofilesmissingnotatrandomADMMlow-rank
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

Recommender systems need matrix completion methods that work with discrete star ratings, survive fake-user manipulation, and handle ratings that are missing not at random. This paper proposes robust discrete matrix completion (RDMC), which fits predictions to the rating scale while using a robust loss that downweights corrupted observations and a separate continuous low-rank matrix that carries the underlying structure. The central empirical claim is that under the three standard nuke attack schemes—average, reverse bandwagon, and love/hate—RDMC keeps the mean prediction shift on the target item close to zero, whereas Soft-Impute shifts by a full rating category or more. In attack-free settings RDMC pays only a small accuracy price compared with discretized Soft-Impute, and it often matches or beats it. The paper also offers its simulation and case-study protocol as a reproducible template for evaluating recommender methods under realistic conditions.

What carries the argument

The engine is an alternating direction method of multipliers over two coupled matrices: a discrete matrix whose entries are restricted to the column-specific rating categories, and a continuous auxiliary matrix that absorbs the nuclear-norm low-rank regularization. Each iteration alternates a soft-thresholded singular value decomposition for the continuous matrix with an elementwise categorical argmin for the discrete matrix, then updates the discrepancy multiplier and penalty. The robust loss—pseudo-Huber, absolute, or truncated absolute—replaces the squared error on observed cells so that fake profiles with extreme ratings do not dominate the fit, and the same loss is used in repeated holdout validation to choose the regularization parameter.

What would settle it

Re-run the simulation with a push attack that tries to promote an unpopular item, or with fake profiles that know the RDMC model's parameters; if the mean prediction shift on the attacked item becomes as large as Soft-Impute's, the claimed reliability advantage does not generalize to that attack class.

Watch

Extended reading notes

Core claim

The central claim is that the combination of a robust loss, a discreteness constraint, and a low-rank constraint in one objective is enough to keep a recommender system stable under profile-injection attacks while staying competitive when no attack occurs. RDMC minimizes a robust loss on the observed rating residuals subject to predictions lying on the rating categories, with a nuclear-norm-regularized continuous auxiliary matrix kept close to the discrete solution through an augmented Lagrangian. In the simulations, RDMC's mean prediction shift stays near zero across all three nuke attack schemes and both missing-data mechanisms, while Soft-Impute's predictions for the target item drop by at least one rating category in many settings. On MovieLens 100K the mean prediction shift of RDMC is close to zero for attack sizes up to 20 percent of the target item's observed ratings, and on the Yahoo! Music data RDMC has the lowest test-set mean absolute error under both missing-not-at-random and missing-completely-at-random mechanisms.

Load-bearing premise

The robustness evidence rests on three specific fake-profile schemes that demote a single popular item, so if real attackers use other strategies the protection shown here may not carry over.

Editorial extensions

If this is right

  • Under average, reverse bandwagon, and love/hate attacks, RDMC keeps the target item's mean prediction shift close to zero, while Soft-Impute shifts by a full rating category or more in many settings.
  • Discretizing Soft-Impute's continuous predictions improves its attack-free accuracy, especially with five or fewer rating categories, but leaves it as vulnerable to fake profiles as the original.
  • A liberal stopping criterion of about ten iterations can reduce RDMC's computation time by more than a factor of three with little loss in prediction accuracy, except under love/hate attacks where the strict criterion is better.
  • RDMC's accuracy is competitive with or better than Soft-Impute variants on real data under both missing-not-at-random and missing-completely-at-random missingness.
  • The same robust discrete completion procedure shows preliminary promise for imputing rating-scale survey data with careless or bot respondents.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • The paper's attack generator covers nuke attacks against one popular item; a natural stress test is to repeat the protocol with push attacks, attacks on several items, or model-aware attacks to see whether the robustness generalizes.
  • Mean prediction shift is an average over missing target cells; an implicit assumption is that this tracks how much user-facing recommendations change, so a sharper test would examine ranking changes or simulated click-through.
  • In the survey simulations the truncated absolute loss is the best choice when many respondents are careless, suggesting the loss function should be tuned to the expected type and concentration of corruption rather than fixed once.
  • The discreteness constraint could also be exploited in strategic missing-data settings, where withholding unfavorable values is the manipulation; the paper sketches this direction but does not test it.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. The paper proposes Robust Discrete Matrix Completion (RDMC), an ADMM-based matrix completion method for discrete rating-scale data that combines a robust loss on observed ratings, a discreteness constraint on predictions, and a nuclear-norm low-rank constraint on an ancillary continuous matrix. Rating columns are centered by their observed median. The method is evaluated against Soft-Impute, a discretized Soft-Impute variant, and simple imputation benchmarks in simulations and two case studies (MovieLens 100K, Yahoo! Music), under nuke attacks and MNAR/MCAR missingness. The authors report that RDMC is substantially more stable than Soft-Impute under attacks, with mean prediction shifts near zero, while paying a small accuracy price in the absence of attacks. A CRAN package RMCLab is provided, and the paper proposes the simulation design as a reproducibility blueprint.

Significance. If the central claim is confirmed, the paper makes a useful contribution: it jointly addresses discreteness, robustness to fake profiles, and MNAR missingness, and it provides a concrete, reproducible evaluation design with public code. The method is clearly described, the regularization parameter is selected by repeated holdout validation, and the robust loss tuning parameters are tied to the rating scale rather than to evaluation outcomes. The empirical work is extensive: simulations across K=3,5,10 categories and two real datasets. However, the main robustness result is currently confounded by a preprocessing difference between RDMC and Soft-Impute, and the no-attack accuracy claim is not uniform across rating scales. These issues are fixable and do not undermine the algorithmic contribution itself, but they determine whether the stated reliability mechanism is actually identified.

major comments (3)
  1. [§2.3 and §3.4, Figures 2 and 5] The comparison underlying the central robustness claim is confounded by the choice of column centering. RDMC centers each column by the observed median and initializes L by median imputation (Section 2.3), whereas Soft-Impute is run on mean-centered training data (Section 3.4). In the nuke attacks, ε=0.2 of the target item's observed ratings are replaced by the minimum rating (Section 3.3); this shifts the target column mean but leaves the median unchanged at contamination below 50%. The large negative MPS of SI and SI-discretized in Figures 2 and 5 can therefore be driven by the centering step alone, before the robust loss or discreteness/low-rank coupling plays any role, and indeed plain median imputation is itself stable for K=3 and K=5 in Figure 2. No ablation varying centering while holding the method fixed is reported, so the paper's Section 5 claim that RDMC's robustness stems from its robust loss and discreteness constraint is not identified. Please add RDMC with mean centering and SI/median imputation with median centering, or otherwise isolate the centering contribution.
  2. [§3.3 and §5] The robustness evidence covers only nuke attacks aimed at demoting a single popular target item, with attack size ε=0.2 in simulations and ε∈{0.10,0.15,0.20} in MovieLens, and MPS measured on missing cells of that target item. The concluding claim that RDMC 'protects well against adversarial manipulation' is therefore broader than the evidence: push or bandwagon attacks, attacks on multiple items, model-aware attacks, and metrics that track actual recommendation harm (e.g., ranking displacement or exposure) are not considered. This does not invalidate the method, but the scope of the reliability claim should be stated as applying to the studied nuke attack family and target-item definition, or additional attack settings should be reported.
  3. [§5 and Figure 1] The statement that RDMC pays 'only a small price' in the absence of corrupted observations is not uniformly supported by Figure 1. With K=3 rating categories, RDMC-pHuber is close to median-discretized imputation and clearly worse than SI-discretized; with K=5 in the MNAR setting it lies between SI and SI-discretized; only for K=10 does it dominate both SI variants. The discussion in Section 5 should qualify the no-attack comparison by the number of rating categories and should not present the price as uniformly small.
minor comments (4)
  1. [§3.6] The text says 'median regression is stable for three and five rating categories'; this should be 'median imputation' or 'median-discretized imputation', and the sentence should state which benchmark is being referred to.
  2. [§2.2] The convergence criterion is a relative change in the objective function, but for the nonconvex truncated absolute loss there is no convergence guarantee; the paper states that tmax=100 sufficed in all experiments, yet it should explicitly acknowledge the lack of a guarantee and report how often the iteration limit was reached.
  3. [§3.6 and Figure A.3] The text says the MCAR attack results are 'qualitatively similar' to MNAR, but Figure A.3 shows a small number of large negative MPS instances for RDMC that are not present in the MNAR setting; this difference should be acknowledged in the main text rather than only in the appendix.
  4. [Data availability] The replication files are promised to be made public upon acceptance, but for a reproducibility-focused paper they should be available at review time; please provide a repository link or an anonymized copy in the current version.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: RDMC's robustness claims are empirical comparisons against external benchmarks, with no fitted input relabeled as prediction.

full rationale

No significant circularity. RDMC's construction (Eq. 2) is an explicit optimization problem with a robust loss, a discreteness constraint on L, and a nuclear-norm penalty on an ancillary matrix Z; the ADMM updates in Eqs. (3)-(4) follow from standard soft-thresholded SVD and elementwise minimization. The regularization parameter lambda is selected by repeated holdout validation on held-out rating cells (Section 2.5), and the robust-loss parameters tau=1 and tau=(K-1)/2 are fixed by the rating scale rather than fitted to the MAE or MPS evaluation outcomes. The central robustness claims are empirical comparisons against external benchmarks (Soft-Impute, median imputation, mode imputation) on simulated and real data, not derivations from the method's own fitted values. Self-citations, such as Alfons and Welz (2024) and Welz and Alfons (2025) for the pseudo-Huber loss and the RMCLab package, are motivational or reproducibility-related and do not carry the argument. The median-versus-mean centering difference (Sections 2.3 versus 3.4) is a legitimate concern about attributing the observed MPS gap to the robust loss rather than to preprocessing, but it is an experimental confound, not a circular step: no equation equates RDMC's MPS to the median center, and no fitted parameter is renamed as a prediction. The derivation chain is therefore not equivalent to its inputs.

Assumptions & free parameters 4 free parameters · 4 assumptions · 0 invented entities

The method introduces no new physical or conceptual entities beyond the mathematical variables L, Z, and Theta used in the optimization. The free parameters are the nuclear-norm penalty lambda, the ADMM schedule, and the robust-loss scale parameters. The key assumptions are the low-rank discrete rating model, the representativeness of the three nuke attacks, and empirical rather than proven convergence of ADMM for the nonconvex loss.

free parameters (4)
  • Regularization parameter lambda = Selected from a logarithmic grid of ten values between 0.01 and 1, scaled by the largest singular value of the…
    Chosen by repeated holdout validation on held-out rating cells; the central predictions depend on this choice.
  • ADMM schedule parameters mu0 and delta = mu0 = 0.1, delta = 1.05
    Set by hand following Huang et al. (2013) and Tang and Guan (2020); not tuned to the evaluation data.
  • Pseudo-Huber scale tau = tau = 1
    Fixed to the step size between rating categories; a modeling choice that controls downweighting strength.
  • Truncated absolute loss threshold tau = tau = (K-1)/2
    Fixed to half the range of the rating scale; a modeling choice that controls the truncation point.
assumptions (4)
  • standard math The nuclear norm is an adequate convex relaxation of the low-rank constraint for discrete rating matrices.
    Inherited from the matrix completion literature and used in Equation (2); not derived in this paper.
  • domain assumption Observed ratings are generated from a latent low-rank continuous signal that is then discretized.
    Used implicitly by the method and explicitly in the simulation design of Section 3.1; if the true rating matrix is not approximately low-rank, the method loses its basis.
  • domain assumption The three nuke attack schemes and the mean prediction shift metric capture practically relevant manipulation.
    The central robustness claim is evaluated only under these attacks and this metric, as described in Sections 3.3 and 3.5.
  • ad hoc to paper ADMM with the nonconvex truncated absolute loss converges to a useful solution within tmax = 100 iterations.
    No convergence proof is given; the claim is based on the authors' numerical experiments in Sections 2.2 and 2.4.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Towards Reliable Recommender Systems for Rating Data." pith.science (2026). https://pith.science/paper/HRJECL4O

@misc{pith2026241220802,
  author       = {Pith},
  title        = {Pith review of: Towards Reliable Recommender Systems for Rating Data},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HRJECL4O}},
  note         = {Machine review of arXiv:2412.20802}
}
read the original abstract

Recommender systems are widely used in the digital landscape to match users with content fitting their preferences. However, growing concerns about fake accounts, strategic manipulation, and other deceptive online behavior place increasing pressure on the reliability of these systems. A common statistical approach behind recommender systems is so-called matrix completion, which predicts how users would rate items they have not yet consumed based on patterns in observed ratings. Realistically applying matrix completion methods requires jointly addressing several overlooked challenges: (i) ratings on discrete scales (such as 1--5 stars); (ii) the presence of malicious users who deliberately manipulate the system to their advantage through fake profiles; (iii) ratings missing not at random since users are more likely to consume items they expect to like; and (iv) fostering transparency, reproducibility, and stability. We jointly address these challenges by proposing a novel method, Robust Discrete Matrix Completion (RDMC), designed to capture the key characteristics of sparse rating data while remaining reliable in the presence of manipulation. We evaluate RDMC through two case studies and carefully designed simulation experiments. Our work thereby offers a statistically-sound blueprint for future studies on how to evaluate recommender systems under realistic scenarios.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

70 extracted references · 46 canonical work pages

  1. [1]

    & McSherry, F

    Achlioptas, D. & McSherry, F. (2001). Fast computation of low rank matrix approximations. STOC '01: Proceedings of the Thirty-Third Annual ACM Symposium on Theory of Computing , 611–618. https://doi.org/10.1145/380752.380858

    work page arXiv 2001

  2. [2]

    Adamopoulos, P. (2024). The spillover effect of fraudulent reviews on product recommendations. Management Science , 70(12), 8818--8832. https://doi.org/10.1287/mnsc.2021.02612

    work page arXiv 2024

  3. [3]

    & Archimbaud, A

    Alfons, A. & Archimbaud, A. (2025). RMCLab : Lab for Matrix Completion and Imputation of Discrete Rating Data . R package version 0.1.0. https://CRAN.R-project.org/package=RMCLab

    work page 2025

  4. [4]

    Alfons, A., Ate s , N., & Groenen, P. (2022). A robust bootstrap test for mediation analysis. Organizational Research Methods , 25(3), 591--617. https://doi.org/10.1177/1094428121999096

    work page doi:10.1177/1094428121999096 2022

  5. [5]

    & Welz, M

    Alfons, A. & Welz, M. (2024). Open science perspectives on machine learning for the identification of careless responding: A new hope or phantom menace? Social and Personality Psychology Compass , 18(2), e12941. https://doi.org/10.1111/spc3.12941

    work page doi:10.1111/spc3.12941 2024

  6. [6]

    B., Garrido, L., Jenaro, C., Martinez-Molina, A., & Arias, B

    Arias, V. B., Garrido, L., Jenaro, C., Martinez-Molina, A., & Arias, B. (2020). A little garbage in, lots of garbage out: Assessing the impact of careless responding in personality survey data. Behavior Research Methods , 52(6), 2489--2505. https://doi.org/https://doi.org/10.3758/s13428-020-01401-8

    work page doi:10.3758/s13428-020-01401-8 2020

  7. [8]

    & Ronchetti, E

    Avella Medina, M. & Ronchetti, E. (2015). Robust statistics: A selective overview and new directions. WIREs Computational Statistics , 7(6), 372--393. https://doi.org/10.1002/wics.1363

    work page doi:10.1002/wics.1363 2015

  8. [9]

    Bai, J. & Ng, S. (2021). Matrix completion, counterfactuals, and factor analysis of missing data. Journal of the American Statistical Association , 116(536), 1746--1763. https://doi.org/10.1080/01621459.2021.1967163

    arXiv 2021

Show all 70 references
  1. [10]

    & Steenkamp, J.-B

    Baumgartner, H. & Steenkamp, J.-B. E. (2001). Response styles in marketing research: A cross-national investigation. Journal of Marketing Research , 38(2), 143--156. https://doi.org/10.1509/jmkr.38.2.143.18840

    2001 doi

  2. [11]

    E., Robertson, M

    Becker, T. E., Robertson, M. M., & Vandenberg, R. J. (2019). Nonlinear transformations in organizational research: Possible problems and potential solutions. Organizational Research Methods , 22(4), 831--866. https://doi.org/10.1177/1094428118775205

    2019 doi

  3. [12]

    & Lanning, S

    Bennett, J. & Lanning, S. (2007). The Netflix prize. Proceedings of KDD Cup and Workshop

    2007

  4. [13]

    Bertsimas, D. & Li, M. L. (2023). Interpretable matrix completion: A discrete optimization approach. INFORMS Journal on Computing , 35(5), 952--965. https://doi.org/10.1287/ijoc.2022.0022

    2023

  5. [14]

    Bond, T. N. & Lang, K. (2019). The sad truth about happiness scales. Journal of Political Economy , 127(4), 1629--1640. https://doi.org/10.1086/701679

    2019 doi

  6. [15]

    Boyd, S., Parikh, N., Chu, E., Peleato, B., Eckstein, J., et al. (2011). Distributed optimization and statistical learning via the alternating direction method of multipliers. Foundations and Trends in Machine Learning , 3(1), 1--122. https://doi.org/10.1561/2200000016

    2011 doi

  7. [16]

    J., & Shen, Z

    Cai, J.-F., Cand \`e s, E. J., & Shen, Z. (2010). A singular value thresholding algorithm for matrix completion. SIAM Journal on Optimization , 20(4), 1956--1982. https://doi.org/10.1137/080738970

    2010 doi

  8. [17]

    & Absil, P.-A

    Cambier, L. & Absil, P.-A. (2016). Robust low-rank matrix completion by Riemannian optimization. SIAM Journal on Scientific Computing , 38(5), S440--S460. https://doi.org/10.1137/15M1025153

    2016 doi

  9. [18]

    & Recht, B

    Cand \`e s, E. & Recht, B. (2009). Exact matrix completion via convex optimization. Foundations of Computational Mathematics , 9(6), 717--772. https://doi.org/10.1007/s10208-009-9045-5

    2009 doi

  10. [19]

    Cand \`e s, E. J. & Tao, T. (2010). The power of convex relaxation: Near-optimal matrix completion. IEEE Transactions on Information Theory , 56(5), 2053--2080. https://doi.org/10.1109/TIT.2010.2044061

    2010

  11. [20]

    Chatterjee, S. (2015). Matrix estimation by universal singular value thresholding. The Annals of Statistics , 43(1), 177--214. https://doi.org/10.1214/14-AOS1272

    2015 doi

  12. [21]

    Chen, Y., Jalali, A., Sanghavi, S., & Caramanis, C. (2013). Low-rank matrix recovery from errors and erasures. IEEE Transactions on Information Theory , 59(7), 4324--4337. https://doi.org/10.1109/TIT.2013.2249572

    2013

  13. [23]

    Cho, J., Kim, D., & Rohe, K. (2019). Intelligent initialization and adaptive thresholding for iterative matrix completion: Some statistical and algorithmic theory for adaptive-impute. Journal of Computational and Graphical Statistics , 28(2), 323--333. https://doi.org/10.1080/...

    2019

  14. [25]

    CMA to investigate Amazon and Google over fake reviews

    Competition and Markets Authority (2021). CMA to investigate Amazon and Google over fake reviews . https://www.gov.uk/government/news/cma-to-investigate-amazon-and-google-over-fake-reviews. Accessed May 26, 2025

    2021

  15. [26]

    Credé, M. (2010). Random responding as a threat to the validity of effect size estimates in correlational research. Educational and Psychological Measurement , 70(4), 596--612. https://doi.org/10.1177/0013164410366686

    2010 doi

  16. [27]

    Davenport, M. A. & Romberg, J. (2016). An overview of low-rank matrix recovery from incomplete observations. IEEE Journal of Selected Topics in Signal Processing , 10(4), 608--622. https://doi.org/10.1109/JSTSP.2016.2539100

    2016

  17. [28]

    & van de Geer, S

    Elsener, A. & van de Geer, S. (2018). Robust low-rank matrix estimation. The Annals of Statistics , 46(6B), 3481--3509. https://doi.org/10.1214/17-AOS1666

    2018 doi

  18. [29]

    Digital services act: Commission welcomes political agreement on rules ensuring a safe and accountable online environment

    European Commission (2022). Digital services act: Commission welcomes political agreement on rules ensuring a safe and accountable online environment . https://ec.europa.eu/commission/presscorner/detail/en/ip_22_2545. Accessed May 26, 2025

    2022

  19. [30]

    FTC brings first case challenging fake paid reviews on an independent retail website

    Federal Trade Commission (2019). FTC brings first case challenging fake paid reviews on an independent retail website . https://www.ftc.gov/news-events/news/press-releases/2019/02/ftc-brings-first-case-challenging-fake-paid-reviews-independent-retail-website. Accessed May 26, 2025

    2019

  20. [31]

    Trade regulation rule on the use of consumer reviews and testimonials - 16 CFR part 465 - RIN 3084–AB76

    Federal Trade Commission (2024). Trade regulation rule on the use of consumer reviews and testimonials - 16 CFR part 465 - RIN 3084–AB76 . Federal Register, 89(163), 68034--68079. https://www.federalregister.gov/documents/2024/08/22/2024-18519/trade-regulation-rule-on-the-use-...

    2024

  21. [32]

    Gunes, I., Kaleli, C., Bilge, A., & Polat, H. (2014). Shilling attacks against recommender systems: A comprehensive survey. Artificial Intelligence Review , 42(4), 767--799. https://doi.org/10.1007/s10462-012-9364-9

    2014 doi

  22. [33]

    Harper, F. M. & Konstan, J. A. (2015). The MovieLens datasets: History and context. ACM Transactions on Interactive Intelligent Systems (TiiS) , 5(4), 19. https://doi.org/10.1145/2827872

    2015 doi

  23. [34]

    D., & Zadeh, R

    Hastie, T., Mazumder, R., Lee, J. D., & Zadeh, R. (2015). Matrix completion and low-rank SVD via fast alternating least squares. Journal of Machine Learning Research , 16(104), 3367--3402

    2015

  24. [35]

    Huang, J., Nie, F., & Huang, H. (2013). Robust discrete matrix completion. Proceedings of the AAAI Conference on Artificial Intelligence , volume 27, 424--430. https://doi.org/10.1609/aaai.v27i1.8675

    2013 doi

  25. [36]

    Huo, Z., Liu, J., & Huang, H. (2016). Optimal discrete matrix completion. Proceedings of the AAAI Conference on Artificial Intelligence , volume 30, 1687--1693. https://doi.org/10.1609/aaai.v30i1.10300

    2016 doi

  26. [37]

    Iimori, H., de Abreu, G. T. F., Taghizadeh, O., & Ishibashi, K. (2020). Discrete-aware matrix completion via proximal gradient. 2020 54th Asilomar Conference on Signals, Systems, and Computers , 1327--1332. https://doi.org/10.1109/IEEECONF51394.2020.9443404

    2020

  27. [38]

    Lam, S. K. & Riedl, J. (2004). Shilling recommender systems for fun and profit. Proceedings of the 13th International Conference on World Wide Web , 393--402. https://doi.org/10.1145/988672.988726

    2004

  28. [39]

    M., Banks, D., Fu, L., Li, M., Tang, Z., & Wu, Q

    LeBlanc, P. M., Banks, D., Fu, L., Li, M., Tang, Z., & Wu, Q. (2024). Recommender systems: A review. Journal of the American Statistical Association , 119(545), 773--785. https://doi.org/10.1080/01621459.2023.2279695

    2024

  29. [40]

    & Zhu, D

    Lee, J.-S. & Zhu, D. (2012). Shilling attack detection—a new approach for a trustworthy recommender system. INFORMS Journal on Computing , 24(1), 117--131. https://doi.org/10.1287/ijoc.1100.0440

    2012

  30. [41]

    & Zhou, D.-X

    Lei, Y. & Zhou, D.-X. (2019). Analysis of singular value thresholding algorithm for matrix completion. Journal of Fourier Analysis and Applications , 25(6), 2957--2972. https://doi.org/10.1007/s00041-019-09688-8

    2019 doi

  31. [42]

    Little, R. J. & Rubin, D. B. (2019). Statistical Analysis with Missing Data . John Wiley & Sons

    2019

  32. [43]

    X., & Wong, R

    Mao, X., Chen, S. X., & Wong, R. K. (2019). Matrix completion with covariate information. Journal of the American Statistical Association , 114(525), 198--210. https://doi.org/10.1080/01621459.2017.1389740

    2019

  33. [44]

    Mazumder, R., Hastie, T., & Tibshirani, R. (2010). Spectral regularization algorithms for learning large incomplete matrices. Journal of Machine Learning Research , 11(80), 2287--2322

    2010

  34. [45]

    Mazumder, R., Saldana, D., & Weng, H. (2020). Matrix completion with nonconvex regularization: Spectral operators and scalable algorithms. Statistics and Computing , 30(4), 1113--1138. https://doi.org/10.1007/s11222-020-09939-5

    2020 doi

  35. [46]

    Mobasher, B., Burke, R., Bhaumik, R., & Williams, C. (2007). Toward trustworthy recommender systems: An analysis of attack models and algorithm robustness. ACM Transactions on Internet Technology (TOIT) , 7(4), 23--es. https://doi.org/10.1145/1278366.1278372

    2007

  36. [47]

    & Fazel, M

    Mohan, K. & Fazel, M. (2012). Iterative reweighted algorithms for matrix rank minimization. Journal of Machine Learning Research , 13(110), 3441--3473

    2012

  37. [48]

    M., Tsiligianni, E., & Deligiannis, N

    Nguyen, D. M., Tsiligianni, E., & Deligiannis, N. (2018). Learning discrete matrix factorization models. IEEE Signal Processing Letters , 25(5), 720--724. https://doi.org/10.1109/LSP.2018.2823268

    2018

  38. [49]

    T., Kim, J., & Shim, B

    Nguyen, L. T., Kim, J., & Shim, B. (2019). Low-rank matrix completion: A contemporary survey. IEEE Access , 7, 94215--94237. https://doi.org/10.1109/ACCESS.2019.2928130

    2019

  39. [50]

    Nie, F., Wang, H., Huang, H., & Ding, C. (2015). Joint S chatten p -norm and _p -norm robust matrix completion for missing value recovery. Knowledge and Information Systems , 42(3), 525--544. https://doi.org/10.1007/s10115-013-0713-z

    2015 doi

  40. [51]

    Peng, J., Hahn, J., & Huang, K.-W. (2023). Handling missing values in information systems research: A review of methods and assumptions. Information Systems Research , 34(1), 5--26. https://doi.org/10.1287/isre.2022.1104

    2023

  41. [52]

    R : A Language and Environment for Statistical Computing

    R Core Team (2025). R : A Language and Environment for Statistical Computing . R Foundation for Statistical Computing, Vienna, Austria. https://www.R-project.org/

    2025

  42. [53]

    Rennie, J. D. & Srebro, N. (2005). Fast maximum margin matrix factorization for collaborative prediction. ICML '05: Proceedings of the 22nd International Conference on Machine Learning , 713--719. https://doi.org/10.1145/1102351.1102441

    2005

  43. [54]

    Ruppel, F., Muma, M., & Zoubir, A. M. (2020). Globally optimal robust matrix completion based on M -estimation. 2020 IEEE 30th International Workshop on Machine Learning for Signal Processing (MLSP) , 1--6. https://doi.org/10.1109/MLSP49062.2020.9231573

    2020

  44. [55]

    Schiavon, L., Nipoti, B., & Canale, A. (2024). Accelerated structured matrix factorization. Journal of Computational and Graphical Statistics , 33(3), 917--927. https://doi.org/10.1080/10618600.2023.2301072

    2024

  45. [56]

    & Kong, L

    Shang, P. & Kong, L. (2021). Regularization parameter selection for the low rank matrix recovery. Journal of Optimization Theory and Applications , 189(3), 772--792. https://doi.org/10.1007/s10957-021-01852-9

    2021 doi

  46. [57]

    Si, M. & Li, Q. (2020). Shilling attacks against collaborative recommender systems: A review. Artificial Intelligence Review , 53(1), 291--319. https://doi.org/10.1007/s10462-018-9655-x

    2020 doi

  47. [58]

    & Jaakkola, T

    Srebro, N. & Jaakkola, T. (2003). Weighted low-rank approximations. ICML '03: Proceedings of the 20th International Conference on Machine Learning , 720--727

    2003

  48. [59]

    Srebro, N., Rennie, J., & Jaakkola, T. (2004). Maximum-margin matrix factorization. Advances in Neural Information Processing Systems , volume 17

    2004

  49. [60]

    Storozuk, A., Ashley, M., Delage, V., & Maloney, E. A. (2020). Got bots? Practical recommendations to protect online survey data from bot attacks. The Quantitative Methods for Psychology , 16(5), 472--481. https://doi.org/10.20982/tqmp.16.5.p472

    2020 doi

  50. [61]

    & Guan, W

    Tang, L. & Guan, W. (2020). Robust matrix completion with complex noise. Multimedia Tools and Applications , 79(3--4), 2703--2717. https://doi.org/https://doi.org/10.1007/s11042-019-08430-2

    2020 doi

  51. [62]

    & Tanaka, M

    Tatsukawa, M. & Tanaka, M. (2018). Box constrained low-rank matrix approximation with missing values. Proceedings of the 7th International Conference on Operations Research and Enterprise Systems - ICORES , 78--84. https://doi.org/10.5220/0006612100780084

    2018 doi

  52. [63]

    Turk, A. M. & Bilge, A. (2019). Robustness analysis of multi-criteria collaborative filtering algorithms against shilling attacks. Expert Systems with Applications , 115, 386--402. https://doi.org/10.1016/j.eswa.2018.08.001

    2019 doi

  53. [64]

    & Yan, X

    Van Roy, B. & Yan, X. (2010). Manipulation robustness of collaborative filtering. Management Science , 56(11), 1911--1929. https://doi.org/10.1287/mnsc.1100.1232

    2010

  54. [65]

    & Fan, J

    Wang, B. & Fan, J. (2025). Robust matrix completion with heavy-tailed noise. Journal of the American Statistical Association , 120(550), 922--934. https://doi.org/10.1080/01621459.2024.2375037

    2025

  55. [66]

    & Meade, A

    Ward, M. & Meade, A. W. (2023). Dealing with careless responding in survey data: Prevention, identification, and recommended best practices. Annual Review of Psychology , 74(1), 577--596. https://doi.org/10.1146/annurev-psych-040422-045007

    2023 doi

  56. [67]

    & Alfons, A

    Welz, M. & Alfons, A. (2025). When respondents don't care anymore: Identifying the onset of careless responding . arXiv :2303.07167. https://doi.org/10.48550/arXiv.2303.07167

    2025 doi

  57. [68]

    Welz, M., Archimbaud, A., & Alfons, A. (2024a). How much carelessness is too much? Q uantifying the impact of careless responding . PsyArXiv. https://doi.org/10.31234/osf.io/8fj6p

    2024 doi

  58. [69]

    Welz, M., Mair, P., & Alfons, A. (2024b). Robust estimation of polychoric correlation . arXiv:2407.18835. https://doi.org/10.48550/arXiv.2407.18835

    2024 doi

  59. [70]

    Xu, Y., Zhuang, F., Wang, E., Li, C., & Wu, J. (2025). Learning Without Missing - At - Random Prior Propensity - A Generative Approach for Recommender Systems . IEEE Transactions on Knowledge and Data Engineering , 37(2), 754--765. https://doi.org/10.1109/TKDE.2024.3490593

    2025

  60. [71]

    Zhang, F., Lu, Y., Chen, J., Liu, S., & Ling, Z. (2017). Robust collaborative filtering based on non-negative matrix factorization and R_1 -norm. Knowledge-Based Systems , 118, 177--190. https://doi.org/10.1016/j.knosys.2016.11.021

    2017 doi

  61. [72]

    & Wang, M

    Zhang, S. & Wang, M. (2019). Correction of corrupted columns through fast robust hankel matrix completion. IEEE Transactions on Signal Processing , 67(10), 2580--2594. https://doi.org/10.1109/TSP.2019.2904021

    2019

  62. [73]

    Zhao, L., Babu, P., & Palomar, D. P. (2016). Efficient algorithms on robust low-rank matrix completion against outliers. IEEE Transactions on Signal Processing , 64(18), 4767--4780. https://doi.org/10.1109/TSP.2016.2572049

    2016

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.