Pith. sign in

REVIEW 3 major objections 5 minor 28 references

The paper claims that coordinated activity can convert into audience-level visibility through recommender feedback loops when rankings reward popularity or feedback signals, and that its Algorithmic Penetration Rate metric captures that con

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

Under controlled simulation, coordinated content reaches non-bot recommendation slots when rankers reward popularity or feedback (APR-Lift up to 0.47 on LastFM), while random ranking shows none.

T0 review reviewed 2026-08-02 challenge →

load-bearing objection A careful simulation study with a useful metric family, but the headline result may be an artifact of the hand-specified response model. the 3 major comments →

arxiv 2607.15114 v2 pith:LRXIMOAV submitted 2026-07-16 cs.IR cs.SI

CoSimRec: Measuring Coordinated-Content Penetration in Recommender Feedback Loops

Nan Li , Jiahong Shao , Jiuyang Lyu This is my paper
classification cs.IR cs.SI
keywords recommender systemsfeedback loopscoordinated behavioragent-based simulationalgorithmic penetration raterecommender robustnessshilling attacks
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The reading

This paper seeks to establish that coordinated online activity—groups of accounts acting together on target content—can pass through a recommender's feedback loop and appear in recommendation lists served to ordinary, non-bot users. It does this with CoSimRec, an offline agent-based simulator in which coordinated accounts, dynamic ranking, controlled non-bot responses, and ranking interventions share one evolving feedback state. The central measure, Algorithmic Penetration Rate (APR), tracks the share of non-bot recommendation slots (exposure APR) and non-bot engagement events (behavior APR) occupied by target content, compared against matched no-attack baselines. In risk-blind experiments, random ranking produced no positive APR-Lift, while popularity-based and feedback-sensitive ranking produced positive lift in all six dataset–recommender settings, reaching 0.4702 on LastFM; a nine-target MovieLens LightGCN stress test turned positive at 25% injection. If correct, the paper demonstrates a computational pathway from organized activity to audience-level exposure and offers a standardized way to measure and intervene on it.

Core claim

The central claim is that coordinated content reaches non-bot recommendation slots through recommender feedback loops when the ranking signal rewards heat or feedback, and that this conversion is measurable. The paper reports that in its controlled simulations, random controls show no statistically supported positive penetration, whereas popularity and feedback ranking produce positive APR-Lift in all six master-worker settings, with LastFM reaching 0.4702 under popularity ranking. A MovieLens 1M LightGCN stress test with nine explicit targets shows all 45 target–seed estimates positive at 25% injection, while no-filler profiles remain near zero. The authors conclude that, under these contro

What carries the argument

The key machinery is the Algorithmic Penetration Rate (APR) metric family, together with the closed-loop simulator CoSimRec. Exposure APR is the fraction of non-bot recommendation slots occupied by target content over the loop; behavior APR is the fraction of non-bot engagement directed to it; APR-Lift compares either to a matched no-attack run, and Penetration Gain normalizes added exposure by the coordinated interaction budget. The simulator couples coordinated accounts, dynamic ranking, controlled non-bot response sampling, and ranking interventions into one shared feedback state, with exposure read directly from served Top-K lists before any user action, so the metric isolates visibility

Load-bearing premise

The load-bearing assumption is that the hand-specified non-bot response model (nine θ weight parameters and fixed heat/freshness update rules), which the paper states is not an empirically calibrated model of clicking, rating, or listening, behaves enough like real users that exposure APR measured from its served lists is meaningful.

What would settle it

Run the identical CoSimRec protocol with the hand-specified response model replaced by a calibrated behavioral model fit to held-out logged interaction data for each dataset; if APR-Lift under popularity or feedback ranking drops to zero or loses significance across seeds, the claim that coordinated activity reaches non-bot slots under realistic responses would be falsified. The paper's own LLM-scorer sensitivity check in the MIND setting—where feedback APR-Lift falls from 0.0838 to 0.0019 (95% CI [-0.0021, 0.0058])—already sharpens this test.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

If this is right

  • Penetration is conditional on ranking objectives: random recommendation yields no positive lift, while popularity- and feedback-based ranking yield positive APR-Lift in all six dataset–recommender settings.
  • Target-rank improvement is not a proxy for audience exposure: LightGCN conditions can show rank promotion without recipient-side APR, so attack-side metrics alone are insufficient for measuring coordinated-content impact.
  • Synchronization-aware ranking interventions reduce APR in every setting tested, with positive intervals in all six dataset–recommender settings, whereas semantic penalties and diversity reranking often show zero or negative APR reduction.
  • Fixed coordinated budgets produce declining APR-Lift as the non-bot audience grows, because the same target exposure spreads across more recommendation slots; this is a property of the fixed-budget design, not evidence of decreasing vulnerability under scaling attacker resources.
  • The protocol provides a reproducible offline standard for comparing coordinated-content penetration across datasets, recommenders, and defenses.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Going beyond the paper: the APR-Lift formulation could be adapted to live serving logs by pairing observed exposure with a counterfactual no-attack baseline estimated from historical traffic, though real deployments lack the simulator's seed-matched control; the main transfer challenge is defining that baseline.
  • Going beyond the paper: the finding that exposure penetration is largest under heat- and feedback-rewarding ranking suggests that platforms can prioritize exposure-level anomaly signals before behavior-level effects, since exposure precedes engagement in the loop.
  • Going beyond the paper: the LightGCN threshold around 25% injection, if it reflects a reachability precondition rather than a volume effect, implies graph connectivity may be the more actionable vulnerability signal for graph-based recommenders; a direct test would vary filler connectivity while holding injection volume fixed.
  • Going beyond the paper: the response-model sensitivity results imply headline APR magnitudes should be read as proof-of-concept under one response specification, not as estimates of real-world prevalence; calibrating the response model to each domain would be needed before using APR as a monitoring metric.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes CoSimRec, an offline agent-based evaluation framework for measuring whether coordinated content activity converts into non-bot exposure and engagement in recommender feedback loops. It introduces the APR metric family (exposure APR, behavior APR, APR-Lift, penetration gain), pairs coordinated runs with matched no-attack baselines, and evaluates on MIND, MovieLens 1M, and LastFM with random, popularity, feedback, MF, BPR-MF, and BPR-LightGCN recommenders. The main empirical claims are that random ranking produces no positive APR-Lift, while popularity and feedback ranking produce positive APR-Lift in all six master-worker settings, and that a nine-target MovieLens LightGCN stress test shows positive mean APR-Lift at 25% injection, with no-filler controls near zero.

Significance. If the results hold, the paper provides a useful measurement protocol for a sociotechnical question that static attack-rank metrics do not capture: whether coordinated inputs can reach non-bot recommendation slots through closed-loop feedback. The strengths are substantial: the protocol uses matched paired seeds, exhaustive two-sided sign-flip tests with Benjamini-Hochberg correction, bootstrap intervals, and a publicly available reproducibility artifact. The paper is also unusually explicit about its limitations, stating in Sections IV-B and VIII that the response model is not an empirically calibrated model of clicking, rating, or listening, and that the LightGCN transition is a five-seed screening result. The contribution is best read as a framework and a controlled demonstration, not as an estimate of real-world attack prevalence. The main open risk is that the primary endpoint is conditional on a hand-specified, uncalibrated non-bot response model, and the paper's own LLM-scorer sensitivity test shows that one plausible response specification collapses the MIND feedback APR-Lift to zero.

major comments (3)
  1. [Section VI-H, Appendix C, Eq. (5), Algorithm 1] The central claim depends on the non-bot response model in Eq. (5) and the feedback-update steps in Algorithm 1 (steps 7 and 9). Exposure APR is read from served lists, but at t>1 those lists are built from a feedback state that includes non-bot actions sampled from Eq. (5). The paper's LLM sensitivity test in Appendix C replaces the deterministic candidate scorer inside Eq. (5) and reduces MIND feedback APR-Lift from 0.0838 to 0.0019 (95% CI [-0.0021, 0.0058]). This is the paper's only scorer-substitution test, and it is confined to one dataset and one recommender. Since the headline positive result disappears in that condition, the claimed computational pathway needs either broader sensitivity analysis across datasets and recommenders, or a substantially more cautious central claim that restricts the conclusion to the single specified response model.
  2. [Eq. (5) and Section IV-B] The response model contains nine behavior weights (theta_m, theta_s, theta_e, theta_q, theta_f, theta_h, theta_i, mu, and the intercept) plus unspecified heat/freshness update rules, and the paper explicitly states that the agents are not calibrated models of clicking, rating, or listening. The parameter values are not printed in the paper; the reader is directed to the artifact. Because the magnitudes and signs of APR-Lift in Table II and Fig. 2 are driven by these choices, the paper should either include the full parameter table and update-rule specification in the main text or appendix, or provide a range/ablation study showing which parameter regions preserve the positive results. Without this, the quantitative APR-Lift values, and especially the 0.4702 LastFM figure, are not independently assessable.
  3. [Section IV-E and Eq. (7)] The popularity-ranking condition is close to constructive by design: the master-worker policy explicitly injects target heat through r_b, mu_sync, and lambda_int, and Eq. (7) directly rewards accumulated heat via beta*H_ct. The positive APR-Lift under popularity ranking therefore mostly demonstrates that injected heat is propagated by a heat-rewarding ranker; it is not independent evidence of a general vulnerability. The paper does note that the policy package rather than synchronization is identified, and that the results are controlled-condition evidence, but the abstract's phrase 'providing evidence of a computational pathway' leans on these settings. I recommend explicitly labeling the popularity condition as a constructive demonstration and anchoring the pathway claim on the feedback-sensitive, latent-factor, and LightGCN results, where the path from injected interactions to non-bot
minor comments (5)
  1. [Section VI-C] The MIND click surrogate has AUROC 0.5925 and AUPRC 0.0581. This is moderate discrimination, but the text already cautions that it is not validation of the response model. Please add a sentence clarifying whether the surrogate was also used in the LLM sensitivity test or only in the separate substitution test, so readers do not conflate the two.
  2. [Section VIII] The limitations section is admirably candid. One small clarification: it states that the LightGCN 25% transition does not generalize beyond the tested conditions. This is useful, but the same caveat should be stated in the abstract, which currently says 'all three target-popularity strata' without the five-seed and single-filler-strategy qualifier.
  3. [Table II] The BH-adjusted p-value of 0.0032 is reported for all six positive tests. Since the minimum exact two-sided p-value with ten seeds is 2/2^10 = 0.001953, it may help to state the rank of the largest p-value in the BH family, or at least note that the reported value is the adjusted value after monotonicity enforcement. This improves reproducibility of the statistical claim.
  4. [Appendix A, Section D] When describing the defense experiments, the paper says MIND includes random bots while MovieLens and LastFM use master-worker coordination. For consistency, please specify whether the random-bot condition uses the same matched baseline and whether the defense conclusions in Fig. 5 aggregate over both attack policies on MIND.
  5. [General] There are several formatting issues in the extracted text (broken equations, missing spaces, e.g., 'TopKc', 'U=\nUN'). These do not affect the science but should be cleaned in the camera-ready version.

Circularity Check

1 steps flagged

Popularity/feedback APR-Lift is largely entailed by the simulator's heat-injection channel; no fitted-parameter or self-citation circularity, but central 'pathway' evidence is partly construction-forced.

specific steps
  1. other [Algorithm 1 steps 3–4; Section IV-D Eq. (7); Section VI-B/Table II]
    "Generate coordinated target interactions B_t ← π_A(U_B, C_A, t). Update feedback state S_t ← Update(S_{t−1}, B_t). ... Popularity ranking responds to accumulated heat ... Score_t(u,c) = αM_uc + βH_ct + γF_ct + δCF_uct + QBonus(c) − λRisk(c)."

    The popularity/feedback APR-Lift is entailed by the simulator's own construction. π_A's purpose is to generate target interactions B_t; Algorithm 1 inserts B_t into the feedback state S_t; Eq. (7) makes the ranking score an explicit function of that state via βH_ct, and the popularity ranker is described as responding to accumulated heat. Hence target heat is raised by the attack and the ranker is defined to reward that exact heat. The headline LastFM value 0.4702 therefore reports the designed heat-injection channel back to the reader; it is a consistency check of the simulator rather than an independent prediction. This is not a fitted-parameter circularity, but the 'computational pathway' evidence is generated by the definition of the loop.

full rationale

No self-citation chain is load-bearing, no fitted parameter is renamed as a prediction, and no uniqueness theorem is imported from the authors' prior work. The APR-Lift values are direct simulation measurements from matched no-attack baselines. The main circularity concern is structural: for popularity and feedback ranking, the coordinated policy is the heat-injection mechanism, the ranker scores by that same heat channel (Eq. 7), and Algorithm 1 explicitly routes coordinated interactions into the feedback state used for ranking. Positive APR-Lift under these rankers is therefore a designed consequence, not an independent empirical discovery. However, the paper's findings are not entirely forced: random-recommender controls remain near zero, and the LightGCN multi-target stress test requires filler-based graph connectivity and retraining to produce positive APR-Lift, so those results carry content beyond the simple heat channel. The LLM scorer-substitution sensitivity (MIND feedback APR-Lift dropping from 0.0838 to 0.0019) is a serious validity threat to the robustness of the central claim, but it is a model-conditional fragility, not a circular derivation; the paper itself labels it as a sensitivity check rather than validation. Overall, the central 'pathway' claim is partially construction-forced in the popularity/feedback conditions, warranting a moderate score, while the LightGCN and random-control results keep the paper from being fully circular.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 3 invented entities

The framework's empirical content rests on a large set of hand-specified parameters (behavior weights, ranker weights, defense weights, attack budgets, engagement weights, risk distributions) whose values are deferred to the reproducibility artifact rather than printed in the paper, plus domain assumptions (the uncalibrated response model, the update rules, the matched-baseline control). The statistics are standard. No new physical entities are postulated; the novel constructs (risk label, agent score) are simulator-internal with no external falsifiable handle.

free parameters (6)
  • Eq. 5 behavior-model weights θ_b, θ_m, θ_s, θ_e, θ_q, θ_f, θ_h, θ_i, μ
    Weights in the Bernoulli response model that determine all non-bot actions, which feed back into rankings (Algorithm 1 step 9). Paper says agents are 'not empirically calibrated' (Section IV-B); values deferred to the artifact.
  • Eq. 7 ranker weights α, β, γ, δ, QBonus, λ
    Weights of the feedback ranker that produce the served Top-K lists (hence exposure APR). Values not printed in the paper.
  • Defense weights λ1-λ4 (Eq. 13)
    Weights of ranking interventions; vary across defense conditions; values in artifact.
  • Coordination policy π_A: r_b, μ_sync, μ_sem, μ_mask, λ_int, b_topic, κ_time = 30-step budgets 552 (random) and 1120 (master-worker) target events
    Attack-package parameters; the paper states comparisons identify the total policy effect, not an isolated synchronization effect (Section IV-E).
  • Engagement weighting (Eq. 10): Click+Like+2Comment+3Share = 1/1/2/3
    Ad-hoc importance weights for behavior APR; no justification given.
  • Target vs organic risk distributions for r_c
    Label-conditioned risk sampled from different target/organic distributions; affects agent score, ranker, and defenses (Section III).
axioms (5)
  • domain assumption Eq. 5's logistic behavior model with hand-set weights is an adequate stand-in for human response in the feedback loop
    The entire loop's dynamics depend on sampled non-bot actions; authors explicitly disclaim calibration ('not empirically calibrated', Section IV-B).
  • domain assumption Heat, freshness, and collaborative-feedback accumulation rules (Eq. 7, popularity mode) faithfully represent recommender update dynamics
    These simulator-defined update rules determine how coordinated activity translates into rank changes.
  • domain assumption Matched no-attack runs with the same target items fully control for candidate-set expansion
    Drives the APR-Lift interpretation; target items are retained in the baseline (Section IV-C).
  • standard math Seed-level sign-flip and paired-bootstrap inference on 10 seeds validly supports the aggregate positive-penetration claim
    Standard nonparametric practice; enumeration of 2^10 sign assignments is correct.
  • standard math Standard probability and statistics (Bernoulli sampling, logistic σ, BH correction)
    Background machinery, uncontroversial.
invented entities (3)
  • Simulator-known risk label r_c no independent evidence
    purpose: Label-conditioned content risk that can enter the agent score, the feedback ranker, and credibility defenses
    Internal simulator construct; no external falsifiable handle; the paper treats risk-blind vs risk-aware conditions as ablations.
  • Profile-conditioned agent score G_uct (Eq. 5) no independent evidence
    purpose: Deterministic candidate-scoring component inside the probabilistic response model
    Simulator-internal scoring; the LLM variant replaces it in sensitivity tests.
  • Coordinated accounts U_B as simulated agents no independent evidence
    purpose: Execute the coordination policy π_A; treated as budget, not as penetration
    Analytical constructs; paper notes policy labels do not imply real accounts are automated (Section IV-E).

reviewed 2026-08-02 · how reviews work

0 comments
Cite this review

Pith. "Pith review of CoSimRec: Measuring Coordinated-Content Penetration in Recommender Feedback Loops." pith.science (2026). https://pith.science/paper/LRXIMOAV

@misc{pith2026260715114,
  author       = {Pith},
  title        = {Pith review of: CoSimRec: Measuring Coordinated-Content Penetration in Recommender Feedback Loops},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/LRXIMOAV}},
  note         = {Machine review of arXiv:2607.15114}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Recommender systems shape which content reaches users, making it important to measure whether coordinated activity gains visibility beyond the accounts that initiate it. Existing robustness evaluations largely focus on static target-rank changes and do not capture how coordinated interactions, recommendation, and user response evolve within a feedback loop. We propose CoSimRec, an offline agent-based evaluation framework that models coordinated accounts, dynamic ranking, controlled non-bot responses, and ranking interventions in a shared closed-loop process. CoSimRec introduces the Algorithmic Penetration Rate (APR) metric family: exposure APR is the primary endpoint, while behavior APR is a response-model-conditional sensitivity measure; both can be compared with matched no-attack baselines. We evaluate CoSimRec on MIND, MovieLens, and LastFM with random, popularity-based, feedback-sensitive, MF, BPR-MF, and BPR-LightGCN recommenders. In a risk-blind primary protocol, random controls show no statistically supported positive penetration, whereas popularity-based and feedback-sensitive ranking produce positive APR-Lift in all six master-worker settings, reaching 0.4702 on LastFM. A nine-target MovieLens 1M LightGCN stress test shows positive mean APR-Lift around 25\% injection in all three target-popularity strata, while no-filler profiles remain near zero. Under these controlled conditions, coordinated inputs reach non-bot recommendation slots, providing evidence of a computational pathway from organized activity to audience-level visibility.

Figures

Figures reproduced from arXiv: 2607.15114 by Jiahong Shao, Jiuyang Lyu, Nan Li.

Figure 1
Figure 1. Figure 1: CoSimRec measures whether coordinated interactions can pass through recommender feedback and reach non-coordinated users. Coordinated accounts [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Risk-blind main penetration evidence from the ten-seed primary artifact. Top: APR-Lift heat map across datasets, recommenders, and coordination [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: MovieLens 1M multi-target LightGCN stress test. Each panel aggregates three explicit targets from one degree stratum. Solid colored lines and shaded [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Scale-up APR-Lift curves under master-worker coordination. Each panel corresponds to one dataset, lines denote recommenders, and shaded bands [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Ten-seed APR-reduction heat map for ranking interventions under master-worker coordination. Rows denote the six dataset–recommender settings, [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

28 extracted references · 2 linked inside Pith

  1. [1]

    Ricci, L

    F. Ricci, L. Rokach, and B. Shapira, Eds.,Recommender Systems Handbook, 3rd ed. Springer, 2022

    2022

  2. [2]

    Matrix factorization techniques for recommender systems,

    Y . Koren, R. Bell, and C. V olinsky, “Matrix factorization techniques for recommender systems,”Computer, vol. 42, no. 8, pp. 30–37, 2009

    2009

  3. [3]

    Bpr: Bayesian personalized ranking from implicit feedback,

    S. Rendle, C. Freudenthaler, Z. Gantner, and L. Schmidt-Thieme, “Bpr: Bayesian personalized ranking from implicit feedback,” inProceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence. AUAI Press, 2009, pp. 452–461

    2009

  4. [4]

    Lightgcn: Simplifying and powering graph convolution network for recommenda- tion,

    X. He, K. Deng, X. Wang, Y . Li, Y . Zhang, and M. Wang, “Lightgcn: Simplifying and powering graph convolution network for recommenda- tion,” inProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval, 2020, pp. 639– 648

    2020

  5. [5]

    The relevance of algorithms,

    T. Gillespie, “The relevance of algorithms,” inMedia Technologies: Essays on Communication, Materiality, and Society, T. Gillespie, P. J. Boczkowski, and K. A. Foot, Eds. Cambridge, MA, USA: MIT Press, 2014, pp. 167–194

    2014

  6. [6]

    Automated media: An institutional theory perspective on algorithmic media production and consumption,

    P. M. Napoli, “Automated media: An institutional theory perspective on algorithmic media production and consumption,”Communication Theory, vol. 24, no. 3, pp. 340–360, 2014

    2014

  7. [7]

    Toward trustwor- thy recommender systems: An analysis of attack models and algorithm robustness,

    B. Mobasher, R. Burke, R. Bhaumik, and C. Williams, “Toward trustwor- thy recommender systems: An analysis of attack models and algorithm robustness,”ACM Transactions on Internet Technology, vol. 7, no. 4, 2007

    2007

  8. [8]

    Shilling attacks against recommender systems: A comprehensive survey,

    I. Gunes, C. Kaleli, A. Bilge, and H. Polat, “Shilling attacks against recommender systems: A comprehensive survey,”Artificial Intelligence Review, vol. 42, pp. 767–799, 2014

    2014

  9. [9]

    Data poisoning attacks on factorization-based collaborative filtering,

    B. Li, Y . Wang, A. Singh, and Y . V orobeychik, “Data poisoning attacks on factorization-based collaborative filtering,” inAdvances in Neural Information Processing Systems, vol. 29. Curran Associates, Inc., 2016, pp. 1885–1893. [Online]. Available: https://proceedings.neurips.cc/paper files/paper/ 2016/hash/83fa5a432ae55c253d0e60dbfa716723-Abstract.html...

    2016

  10. [10]

    How algorithmic confounding in recommendation systems increases homogeneity and decreases utility,

    A. J. B. Chaney, B. M. Stewart, and B. E. Engelhardt, “How algorithmic confounding in recommendation systems increases homogeneity and decreases utility,” inProceedings of the 12th ACM Conference on Recommender Systems, 2018, pp. 224–232

    2018

  11. [11]

    Feedback loop and bias amplification in recommender systems,

    M. Mansoury, H. Abdollahpouri, M. Pechenizkiy, B. Mobasher, and R. Burke, “Feedback loop and bias amplification in recommender systems,” inProceedings of the 29th ACM International Conference on Information and Knowledge Management, 2020, pp. 2145–2148

    2020

  12. [12]

    The unfairness of popularity bias in recommendation,

    H. Abdollahpouri, M. Mansoury, R. Burke, and B. Mobasher, “The unfairness of popularity bias in recommendation,” inProceedings of the Workshop on Recommendation in Multi-Stakeholder Environments, ser. CEUR Workshop Proceedings, vol. 2440, 2019. [Online]. Available: https://ceur-ws.org/V ol-2440/paper4.pdf

    2019

  13. [13]

    The rise of social bots,

    E. Ferrara, O. Varol, C. Davis, F. Menczer, and A. Flammini, “The rise of social bots,”Communications of the ACM, vol. 59, no. 7, pp. 96–104, 2016

    2016

  14. [14]

    Online human-bot interactions: Detection, estimation, and characterization,

    O. Varol, E. Ferrara, C. A. Davis, F. Menczer, and A. Flammini, “Online human-bot interactions: Detection, estimation, and characterization,” in Proceedings of the International AAAI Conference on Web and Social Media, vol. 11, no. 1, 2017, pp. 280–289

    2017

  15. [15]

    The spread of true and false news online,

    S. V osoughi, D. Roy, and S. Aral, “The spread of true and false news online,”Science, vol. 359, no. 6380, pp. 1146–1151, 2018

    2018

  16. [16]

    Generative agents: Interactive simulacra of human behavior,

    J. S. Park, J. C. O’Brien, C. J. Cai, M. R. Morris, P. Liang, and M. S. Bernstein, “Generative agents: Interactive simulacra of human behavior,” inProceedings of the 36th Annual ACM Symposium on User Interface Software and Technology, 2023

    2023

  17. [17]

    User behavior simulation with large language model-based agents,

    L. Wang, J. Zhang, H. Yang, Z. Chen, J. Tang, Z. Zhang, X. Chen, Y . Lin, H. Sun, R. Song, W. X. Zhao, J. Xu, Z. Dou, J. Wang, and J.-R. Wen, “User behavior simulation with large language model-based agents,”ACM Transactions on Information Systems, vol. 43, no. 2, 2025

    2025

  18. [18]

    On generative agents in recommendation,

    A. Zhang, Y . Chen, L. Sheng, X. Wang, and T.-S. Chua, “On generative agents in recommendation,” inProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval, 2024

    2024

  19. [19]

    Recsim: A configurable simulation platform for recommender systems,

    E. Ie, C.-W. Hsu, M. Mladenov, V . Jain, S. Narvekar, J. Wang, R. Wu, and C. Boutilier, “Recsim: A configurable simulation platform for recommender systems,”arXiv preprint arXiv:1909.04847, 2019. [Online]. Available: https://arxiv.org/abs/1909.04847

    Pith/arXiv arXiv 1909

  20. [20]

    Oasis: Open agent social interaction simulations with one million agents,

    Z. Yang, Z. Zhang, Z. Zheng, Y . Jiang, Z. Gan, Z. Wang, Z. Ling, J. Chen, M. Ma, B. Dong, P. Gupta, S. Hu, Z. Yin, G. Li, X. Jia, L. Wang, B. Ghanem, H. Lu, C. Lu, W. Ouyang, Y . Qiao, P. Torr, and J. Shao, “Oasis: Open agent social interaction simulations with one million agents,”arXiv preprint arXiv:2411.11581, 2024. [Online]. Available: https://arxiv....

    Pith/arXiv arXiv 2024

  21. [21]

    Mind: A large-scale dataset for news recommendation,

    F. Wu, Y . Qiao, J.-H. Chen, C. Wu, T. Qi, J. Lian, D. Liu, X. Xie, J. Gao, W. Wu, and M. Zhou, “Mind: A large-scale dataset for news recommendation,” inProceedings of the 58th Annual Meeting of the As- sociation for Computational Linguistics. Association for Computational Linguistics, 2020, pp. 3597–3606

    2020

  22. [22]

    The movielens datasets: History and context,

    F. M. Harper and J. A. Konstan, “The movielens datasets: History and context,”ACM Transactions on Interactive Intelligent Systems, vol. 5, no. 4, pp. 1–19, 2015

    2015

  23. [23]

    Hetrec 2011 datasets,

    GroupLens Research, “Hetrec 2011 datasets,” 2011, dataset release for the 2nd International Workshop on Information Heterogeneity and Fusion in Recommender Systems. [Online]. Available: https: //grouplens.org/datasets/hetrec-2011/

    2011

  24. [24]

    What kind of news gatekeepers do we want machines to be? filter bubbles, fragmentation, and the normative dimensions of algorithmic recommendations,

    E. Nechushtai and S. C. Lewis, “What kind of news gatekeepers do we want machines to be? filter bubbles, fragmentation, and the normative dimensions of algorithmic recommendations,”Computers in Human Behavior, vol. 90, pp. 298–307, 2019

    2019

  25. [25]

    Experimental study of inequality and unpredictability in an artificial cultural market,

    M. J. Salganik, P. S. Dodds, and D. J. Watts, “Experimental study of inequality and unpredictability in an artificial cultural market,”Science, vol. 311, no. 5762, pp. 854–856, 2006

    2006

  26. [26]

    Exposure diversity as a design principle for recommender systems,

    N. Helberger, K. Karppinen, and L. D’Acunto, “Exposure diversity as a design principle for recommender systems,”Information, Communica- tion & Society, vol. 21, no. 2, pp. 191–207, 2018

    2018

  27. [27]

    The logic of connective action: Digital media and the personalization of contentious politics,

    W. L. Bennett and A. Segerberg, “The logic of connective action: Digital media and the personalization of contentious politics,”Information, Communication & Society, vol. 15, no. 5, pp. 739–768, 2012

    2012

  28. [28]

    On the detection of disinformation campaign activity with network analysis,

    L. Vargas, P. Emami, and P. Traynor, “On the detection of disinformation campaign activity with network analysis,” inProceedings of the 2020 ACM SIGSAC Conference on Cloud Computing Security Workshop. ACM, 2020, pp. 133–146. PREPRINT, 2026 11 APPENDIXA SUPPLEMENTARYEXPERIMENTALDETAILS A. Multi-Target BPR-LightGCN Configuration The multi-target stress test...

    2020

This paper was first reviewed by deepseek-v4-flash on August 2, 2026.