Pith. sign in

REVIEW 5 major objections 6 minor 1 cited by

GGBond: Growing Graph-Based AI-Agent Society for Socially-Aware Recommender Simulation

T0 review · 5 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read This paper makes the case that recommender-system evaluation should move from static logs to a living simulated society of cognitive agents, and shows that socially interacting agents reproduce human rating distributions more closely than…

desk verdict Architectural ideas worth taking seriously, but the headline validation experiment is too thin to support them. read the letter →

arxiv 2505.21154 v1 pith:5LR4MUEH submitted 2025-05-27 cs.MA cs.AIcs.CY

Hailin Zhong , Hanlin Wang , Yujun Ye , Meiyi Zhang , Shengxin Zhu This is my paper
classification cs.MAcs.AIcs.CY
keywords AIagentsrecommendersystemssocialsimulationuserbehaviormodelingBigFivepersonalityLLMmulti-agentinfluence
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

GGBond is a simulation platform intended to replace static offline datasets with a living virtual society of users, each modeled as a cognitive agent that remembers past items, has moods, learns preferences, and maintains trust-based social ties. The paper's core argument is that recommender systems should be evaluated under this kind of long-term, socially embedded feedback, not just on historical logs. The central empirical claim is that agents who interact socially for a few rounds produce movie-rating distributions that are much closer to real MovieLens users than the ratings of isolated static agents, with KL divergence dropping from 0.0750 to 0.0108 and Earth Mover's Distance from 0.4200 to 0.0900. If the claim holds, the platform offers a controlled and observable testbed for studying preference drift, social influence, and algorithmic interventions.

What carries the argument

The load-bearing mechanism is the GGBond Graph, a multilayer heterogeneous social network in which each node carries a Big-Five personality vector predicted from structural features, and edges encode interest similarity (Jaccard overlap of liked genres), personality alignment (cosine similarity of trait vectors), and structural homophily (shared demographic attributes). Each agent runs a five-module cognitive loop: episodic memory with exponential forgetting, a valence–arousal affective state that modulates a decision threshold, an exponentially smoothed preference embedding, a social-cognition module that computes intimacy and risk, and the ICR2 motivational engine that produces a scalar drive $C = \alpha I + \beta N + \gamma R - \delta K$ and compares it to an emotion-modulated threshold $\theta = \theta_0 - \kappa V_t$. The agents' watch/rate/share actions write back into memory, emotion, trust, and edge weights, closing a perception–decision–action–feedback loop under a discrete-time scheduler.

What would settle it

Run the rating-consistency experiment twice with identical agent internals, once with the true GGBond social graph and once with a degree-preserving random graph that has the same edge count but shuffled community structure; if the random-graph agents match the human rating distribution as closely as the social-graph agents, the claim that social structure drives human-like ratings is falsified.

Watch

Extended reading notes

Core claim

In the paper's own framing, the discovery is that adding a dynamic social layer to simulated users is what moves their rating behavior measurably closer to human behavior. The GGBond Agent Rating Distribution, produced after three rounds of social exchange in which agents accept or reject neighbors' movie recommendations and update their personality and interest vectors, is compared against the Human Rating Distribution from MovieLens; the reported KL divergence is 0.0108 and the EMD is 0.0900. The Static Agent Rating Distribution, produced by the same initialization but without any social interaction, gives KL = 0.0750 and EMD = 0.4200. The paper interprets this gap as evidence that social dynamics, not just individual personas, shape human-like evaluative tendencies, and it reports that increasing the number of interaction rounds from 0 to 30 improves Recall@20 and NDCG@20 for all three embedded recommender models, with the graph-based LightGCN improving the most.

Load-bearing premise

The load-bearing premise is that the cross-domain mapping from four behavioral features in MovieLens to four structural features in the Facebook graph preserves the same personality meaning, so that degree, neighbor entropy, betweenness, and PageRank can stand in for activity, diversity, conformity deviation, and novelty seeking.

Editorial extensions

If this is right

  • If social interaction is what makes simulated ratings human-like, then recommender evaluations based on isolated users may systematically misjudge how a deployed system will behave.
  • The reported improvements in Recall@20 and NDCG@20 with more interaction rounds imply that dynamically updated profiles, rather than static initial ones, give recommenders a better signal.
  • The interpretable ICR2 decomposition (intimacy, curiosity, reciprocity, risk) allows an experimenter to trace why an agent accepted or skipped an item, making the simulation a diagnostic tool for interventions.
  • Across MF, MultVAE, and LightGCN, the same pattern of rising satisfaction and acceptance rates with interaction depth suggests the social loop itself, not a single recommender, is the driver.

Reading between the lines

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

  • A natural next test the paper does not run is per-user fidelity: comparing the full rating distribution still allows all agents to be wrong in the same way, so a per-user or per-demographic KL/EMD comparison would be a stricter check on social realism.
  • Because the personality-transfer model is trained on MovieLens text and applied via structural proxies to an anonymized Facebook graph, the entire personality layer remains conditional on an untested cross-domain assumption; validating on any node with known traits would settle it.
  • The interaction-depth improvement could be partly a self-reinforcing loop, since agents rate what they are recommended and those ratings become the next round's ground truth; a control in which agents interact but never update their preferences would separate social influence from mere exposure.
  • The paper restricts LLM use to post-decision text generation; letting agents read and react to each other's natural-language reviews would introduce a new channel of social influence that the current numeric intimacy/trust signals do not capture.
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

5 major / 6 minor

Summary. The paper proposes GGBond, an AI-agent simulation platform for recommender systems. It combines five-layer cognitive agents (memory, affect, preference, social cognition, ICR2 motivation, behavior) with a multilayer social graph initialized from the Stanford Facebook topology and augmented with Big-Five personality vectors inferred from MovieLens behavioral features. The main empirical claim is that socially interacting GGBond agents produce rating distributions closer to human MovieLens ratings than static agents (Table I: KL 0.0108 vs 0.0750, EMD 0.0900 vs 0.4200), and that increasing interaction depth monotonically improves recommendation metrics and agent satisfaction. Additional experiments evaluate three recommenders at different interaction depths and report post-recommendation behavioral indicators.

Significance. If validated, the platform would be a useful testbed for long-term recommender evaluation, and the manuscript deserves credit for a detailed modular design, explicit formulas for memory, affect, intimacy, risk, and decision processes, and a clear integration path for three recommendation algorithms. However, the empirical core is thin: the central validation is a single table with two point estimates, no variance or confound controls, a cross-domain personality transfer that is unvalidated, and a behavior-consistency section that cites other papers rather than running experiments. The work is therefore more of a system description than a validated simulation; its significance cannot currently be assessed from the evidence presented.

major comments (5)
  1. [V.A, Table I] The central claim that social interaction improves human-alignment is not supported by the reported evidence. Table I reports only two point estimates per condition, with no number of agents, number of target movies, random seeds, number of independent runs, error bars, or significance tests; the phrase "significantly lower" in the text is therefore not justified. Moreover, the static-agent baseline is not matched on rating concentration: the social-interaction mechanism (Eqs. 20, 29–33) moves preference vectors and thresholds toward neighbor consensus, which mechanically concentrates ratings around the modal 3–4 human scores and lowers KL/EMD relative to an unmoved baseline. The experiment needs a control such as a non-social averaging baseline or a preference-shrinkage-only condition, plus variance reporting, to attribute the improvement to social cognition rather than averaging.
  2. [III.C and V.A] The personality-transfer step is circular with respect to the validation target. The MLP fθ is trained on four behavioral features extracted from MovieLens and is evaluated on a MovieLens hold-out (Section III.C, RMSE < 0.1, Pearson r > 0.6); the same MovieLens distribution is then used as the human ground truth in the rating-consistency experiment of Section V.A. Because the GGBond agents' personalities are derived from MovieLens behavioral statistics, part of the agreement between GRD and HRD is built in by construction, and the structural proxies (degree, neighbor entropy, betweenness, PageRank) are never validated against any external ground truth in the Stanford domain. This unvalidated transfer is load-bearing: the personality vectors feed the personality graph layer (Eq. 14), the intimacy score (Eq. 24), and the risk function (Eq. 26), so all downstream social and trust computations depend on it. A demonstration on a held-out domain or an ablation without the transferred personality layer is required.
  3. [V.B] The Behavior Consistency subsection does not contain any experiments. It cites [45] (and related work) for trust calibration, social influence susceptibility, and rational preference disclosure, and then concludes that "these converging findings substantiate our observation." Outsourced evidence is not evidence about the GGBond implementation; the only in-house empirical support for behavioral realism is Table I. The authors should either run the corresponding behavioral tests on GGBond agents or substantially weaken the claim to a design proposal.
  4. [VI, Tables II and III] The evaluation of interaction depth is also statistically unsupported and confounded. Tables II and III report no error bars, seeds, or significance tests; the improvements are numerically small (e.g., MF Recall@20 0.1502→0.1623), and because the recommender models are trained on agent-generated profiles that converge under social influence, the gains may reflect increased profile homogeneity rather than alignment with human preferences. The behavioral indicators in Table III (Eqs. 39–42) are internal consistency metrics, not comparisons to human behavior, so they do not validate realism. Reporting per-run variance and a non-social convergence baseline would clarify what is being measured.
  5. [IV.D, Eq. 28 and Table I] The motivation engine and the social graph construction contain at least a dozen free parameters (e.g., α, β, γ, δ in Eq. 28; σV, σA in Eq. 19; η in Eq. 20; λmem in Eq. 17; τu and σu in Eqs. 31–32; λdemo, λpref in Eq. 25; ρI, ρR in Eq. 30), and the paper gives no sensitivity analysis. Since the reported rating-consistency numbers are point estimates from a single configuration, it is unclear whether the qualitative outcome is robust or a consequence of parameter tuning. A sensitivity analysis over the main parameters (at least the ICR2 weights and η) is needed to support the claim that the architecture, rather than the chosen defaults, produces human-like distributions.
minor comments (6)
  1. [V.A] The definitions of EMD and KL are generic; please state how the distributions are constructed (e.g., per-movie or aggregate), how ratings are pooled, and how zero-probability bins are handled for KL, since Figure 6 shows a single density plot with no indication of the number of agents or movies.
  2. [III.C] The text reports "Real banchmark datasets" (typo) and the evaluation claim "RMSE < 0.1, Pearson r > 0.6" lacks the number of test users and whether the scores are per-trait or pooled; please clarify.
  3. [References] References [44] and [45] appear to be the same work ("Can large language model agents simulate human trust behavior?") with different author lists; please merge and cite consistently.
  4. [IV.D and V.B] The Social Exchange Theory citation in Section IV.D is unresolved ("[?]"), and the SimUSER citation in Section V.B is also unresolved ("[?]"); please fix these citations.
  5. [VII.B] The related work incorrectly attributes Voyager to DeepSeek-R1; Voyager uses GPT-4 as its backbone, so the sentence should be corrected to avoid a factual error.
  6. [Abstract and IV.D] The acronym "ICR2" is introduced in the abstract and used as "IC2" in Section IV.D; please unify the notation throughout the paper.

Circularity Check

2 steps flagged · score 6.0 of 10

Rating-consistency validation leaks MovieLens-derived personas into the MovieLens target, and the recommendation-evaluation loop defines its own ground truth.

  1. fitted input called prediction [Section III (Personality Prediction Model) and Section V.A (Rating Consistency)]
    "we derive behavioral statistics from user interactions in MovieLens, including activity level, diversity, rating deviation, and novelty preference. … T_conf_u = 1/|I_u| Σ_{i∈I_u} (r_ui − R̄_i)^2 … T_nov_u = 1/|I_u| Σ_{i∈I_u} 1/pop(i) … (1) Human Ratings: Empirical ground-truth ratings are sourced from the MovieLens dataset, where each item (movie) has been rated by a diverse population of real users."

    The agent personas are not independent of the validation target: the MLP that produces Big-Five vectors is trained on MovieLens behavioral features that are direct functions of the MovieLens rating distribution (rating deviation from the global mean, inverse popularity, genre entropy). Those same MovieLens ratings are then used as the human ground truth in Table I. The GGBond ARD therefore inherits the dataset's consensus and popularity structure before any social interaction occurs, so the absolute KL/EMD agreement with HRD is partly built into the initialization rather than earned by the social simulation.

  2. self definitional [Section VI.A (Recommendation Performance), Eqs. 20, 32, 37]
    "the profile data of every agent, including interaction and rating data, are sent into the recommender systems … Each agent evaluates the presented movies (based on content similarity and internal state), and accepts or rejects accordingly. … pu ← (1−η)·pu + η·em … r = r̂u,m + ε … Recall@20 = 1/|V| Σ_{u∈V} |R^{(20)}_u ∩ G_u| / |G_u|, where R^{(20)}_u is the top-20 recommendation list for agent u, and G_u is the set of relevant (positively rated) items by u."

    The 'ground truth' relevant set G_u for Recall/NDCG is the set of items the agent itself positively rates, but those ratings are generated from the same preference vector pu and predicted score r̂u,m that the recommender used to create the top-k list. Eq. 20 moves pu toward the embedding of accepted items, so after each round the recommender and the agent's acceptance criterion converge on the same vectors. The observed monotone increases in Recall@20 and NDCG@20 (Table II) therefore measure self-consistency of a closed loop—recommender output shapes acceptance, acceptance reshapes profiles, profiles shape recommender output—not an independent improvement in prediction quality or user satisfaction. This is an evaluation whose target is defined by its own inputs.

full rationale

The architecture itself is not wholly circular: the GGBond-vs-static rating comparison is a real experiment, and the ICR2 coefficients are hand-set rather than fitted to Table I. However, two load-bearing empirical validations are self-referential. First, the personas used to initialize agents are inferred from MovieLens behavioral statistics—conformity deviation and novelty seeking are direct functions of the same ratings that later serve as the human ground truth—so the absolute KL/EMD agreement of Table I partly measures the initialization, not emergent social cognition. Second, the Section VI recommendation evaluation defines its ground-truth relevant set as the agents' own positively rated items, while those ratings are produced by the same preference vectors the recommenders consume; the monotone Recall/NDCG gains across rounds therefore reflect closed-loop self-consistency, not external predictive improvement. No load-bearing self-citations were found; the behavior-fidelity support outsourced to [45] is an external citation, not a self-citation. These issues are distinct from the separate confound that social averaging could shrink the rating distribution toward the modal 3-4 region; that confound is a correctness risk, not circularity. Overall, the central social-interaction comparison retains some independent content, but the validation chain contains built-in agreement with its own inputs, warranting a partial-circularity score of 6.

Assumptions & free parameters 12 free parameters · 6 assumptions · 0 invented entities

The central claim rests on a large set of hand-set coefficients and unvalidated transfer assumptions. None of these are derived from first principles or fitted with error bars, and no code is released to test alternatives.

free parameters (12)
  • ICR2 motivational weights (alpha, beta, gamma, delta) = 0.40, 0.35, 0.20, 0.25
    Appear in Eq. 28 as the default configuration of the motivation score C. No sensitivity analysis or fitting procedure is reported.
  • Graph layer fusion weights (alpha, beta, gamma) = unconstrained except sum to 1
    Eq. 16 combines interest, personality, and structural layers. The paper reuses the symbol alpha, creating ambiguity with the ICR2 weight.
  • Preference update rate eta = not specified
    Eq. 20 updates the preference vector with exponential smoothing; the numeric rate is left open.
  • Memory forgetting rate lambda_mem = not specified
    Eq. 17 controls episodic decay and feeds into trust and risk estimates.
  • Affective sensitivities sigma_V, sigma_A = not specified
    Eq. 19 updates valence and arousal after each movie event.
  • Decision threshold theta0 and mood sensitivity kappa = not specified
    Eq. 29 makes acceptance depend on a base threshold and current valence.
  • Stochasticity parameters tau_u and sigma_u = not specified
    Eqs. 31-32 set the softmax temperature for watch decisions and the rating noise scale.
  • Tie update rates rho_I, rho_R = not specified
    Eq. 30 adjusts intimacy and reciprocity after each satisfaction signal.
  • Risk function weights and risk_base_u = 0.4, 0.5, 0.4, 0.6, 0.3, 0.8, 0.5, 0.5 plus per-agent base risk
    Eq. 26 defines risk with fixed coefficients and a per-agent base level; coefficients are asserted without derivation.
  • Reciprocity weights (0.6, 0.3, 0.1) = 0.6, 0.3, 0.1
    Eq. 27 combines share history, interest overlap, and personality complementarity.
  • Homophily scalars lambda_demo, lambda_pref = not specified
    Eq. 25 amplifies structural intimacy with demographic and preference similarity.
  • Language novelty penalty lambda_lang = not specified
    Eq. 21 penalizes movies whose language differs from the agent's language.
assumptions (6)
  • domain assumption Big-Five personality traits can be extracted from short text reviews using a pre-trained RoBERTa classifier and treated as ground truth labels.
    Section III.C uses RoBERTa on MovieLens/Steam/AmazonReviews text to label personalities; no human annotation or reliability check is reported.
  • ad hoc to paper A regression model trained on MovieLens behavioral features transfers to structural features of the Stanford Facebook graph.
    Section III.C applies f_theta to degree, neighbor entropy, betweenness, and PageRank with no validation on the target graph because ground-truth personalities are unavailable.
  • domain assumption A linear weighted sum of intimacy, novelty, reciprocity, and risk describes human acceptance decisions.
    Eq. 28 with the default coefficients is asserted from MAUT and 'empirical behavioral studies' without a fitting or ablation.
  • domain assumption Mood-as-Information and Social Exchange Theory justify the threshold rule and tie update rules.
    Eq. 29 and Eq. 30 import these theories; their quantitative forms (linear threshold shift, additive tie updates) are not derived.
  • domain assumption MovieLens rating distributions are a valid human ground truth for agents whose personalities are seeded from MovieLens data.
    Section V.A compares agent rating distributions to the same dataset family used to train the personality model.
  • domain assumption DeepSeek-R1 generates sufficiently human-like language for reviews and sharing messages.
    Module 0 relies on an external LLM; no qualitative or quantitative evaluation of generated text is presented.

how reviews work

0 comments
Cite this review

Pith. "Pith review of GGBond: Growing Graph-Based AI-Agent Society for Socially-Aware Recommender Simulation." pith.science (2026). https://pith.science/paper/5LR4MUEH

@misc{pith2026250521154,
  author       = {Pith},
  title        = {Pith review of: GGBond: Growing Graph-Based AI-Agent Society for Socially-Aware Recommender Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/5LR4MUEH}},
  note         = {Machine review of arXiv:2505.21154}
}
read the original abstract

Current personalized recommender systems predominantly rely on static offline data for algorithm design and evaluation, significantly limiting their ability to capture long-term user preference evolution and social influence dynamics in real-world scenarios. To address this fundamental challenge, we propose a high-fidelity social simulation platform integrating human-like cognitive agents and dynamic social interactions to realistically simulate user behavior evolution under recommendation interventions. Specifically, the system comprises a population of Sim-User Agents, each equipped with a five-layer cognitive architecture that encapsulates key psychological mechanisms, including episodic memory, affective state transitions, adaptive preference learning, and dynamic trust-risk assessments. In particular, we innovatively introduce the Intimacy--Curiosity--Reciprocity--Risk (ICR2) motivational engine grounded in psychological and sociological theories, enabling more realistic user decision-making processes. Furthermore, we construct a multilayer heterogeneous social graph (GGBond Graph) supporting dynamic relational evolution, effectively modeling users' evolving social ties and trust dynamics based on interest similarity, personality alignment, and structural homophily. During system operation, agents autonomously respond to recommendations generated by typical recommender algorithms (e.g., Matrix Factorization, MultVAE, LightGCN), deciding whether to consume, rate, and share content while dynamically updating their internal states and social connections, thereby forming a stable, multi-round feedback loop. This innovative design transcends the limitations of traditional static datasets, providing a controlled, observable environment for evaluating long-term recommender effects.

Figures

Figures reproduced from arXiv: 2505.21154 by the authors.

Figure 1
Figure 1. GGBond System Architecture: Rrecommender Engine, Database, Social network, Agent [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Big-5 Model Training Process Definition III.1 (Activity Level). Activity Level T act u denotes the total number of items rated by user u, capturing their engagement intensity: T act u = X i∈Iu 1, (1) where Iu is the set of items rated by user u. Definition III.2 (Diversity). Diversity T div u measures the en￾tropy of genre distribution among all rated items, reflecting the breadth of a user’s interests: T div u = − … view at source ↗
Figure 3
Figure 3. Big-5 Alignment Framework hold-out set, achieving robust predictive performance (RMSE < 0.1, Pearson r > 0.6). The trained model was subsequently applied to predict the personality traits of Stanford network nodes, producing personality-enriched node representations without reliance on explicit personal data. D. Big-Five Mapping In the Stanford social network dataset, each node v is originally described by a set of … view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: Agent Architecture: Module 0 (GPT4 API), Module 1 (Individual [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Comparison of rating distributions across Human, GGBond Agent, [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 7
Figure 7. Figure 7: Recall@20 across different interaction rounds. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

Discussion (0). Sign in to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Autonomous Information Seeking: A Roadmap for Agentic Recommender Systems

    cs.IR 2026-07 accept novelty 5.0 of 10

    Agentic recommender systems are organized by agent role (assisted, as-recommender, as-simulator) crossed with autonomy levels L2–L5, yielding a roadmap of architectures, evaluation limits, and open challenges.

Reference graph

Works this paper leans on

52 extracted references · 37 canonical work pages · cited by 1 Pith paper

  1. [45]

    Can large language model agents simulate human trust behavior? In Advances in Neural Information Processing Systems (NeurIPS), 2024

    Feiyu Xu et al. Can large language model agents simulate human trust behavior? In Advances in Neural Information Processing Systems (NeurIPS), 2024

    work page 2024

  2. [1]

    Argyle, Ethan C

    Lisa P. Argyle, Ethan C. Busby, Nancy Fulda, Joshua Gubler, Christo- pher Rytting, and David Wingate. Out of one, many: Using language models to simulate human samples. Political Analysis, 31(3):337–351, 2023

    work page 2023

  3. [2]

    Simuser: Simulating user be- havior with large language models for recommender system evaluation

    Nicolas Bougie and Narimasa Watanabe. Simuser: Simulating user be- havior with large language models for recommender system evaluation. arXiv preprint arXiv:2504.12722 , 2024

    arXiv 2024

  4. [3]

    A cooperative species: Human reciprocity and its evolution

    Samuel Bowles and Herbert Gintis. A cooperative species: Human reciprocity and its evolution. In A cooperative species . Princeton University Press, 2011

    work page 2011

  5. [4]

    Agentic feedback loop modeling improves recommendation and user simulation

    Shihao Cai, Jizhi Zhang, Keqin Bao, Chongming Gao, Qifan Wang, Fuli Feng, and Xiangnan He. Agentic feedback loop modeling improves recommendation and user simulation. SIGIR, 2025

    work page 2025

  6. [5]

    Twenty Years of Personality Computing: Threats, Challenges and Future Directions

    Fabio Celli, Aleksandar Kartelj, Miljan Ð or ¯devi´c, Derwin Suhartono, Vladimir Filipovi´c, Veljko Milutinovi´c, Georgios Spathoulas, Alessandro Vinciarelli, Michal Kosinski, and Bruno Lepri. Twenty years of personality computing: Threats, challenges and future directions. arXiv preprint arXiv:2503.02082, 2025

    work page Pith review arXiv 2025

  7. [6]

    Exploring large language model based intelli- gent agents: Definitions, methods, and prospects

    Yuheng Cheng, Ceyao Zhang, Zhengwen Zhang, Xiangrui Meng, Sirui Hong, Wenhao Li, Zihao Wang, Zekai Wang, Feng Yin, Junhua Zhao, and Xiuqiang He. Exploring large language model based intelli- gent agents: Definitions, methods, and prospects. arXiv preprint arXiv:2401.03428, 2024

    arXiv 2024

  8. [7]

    Linton C. Freeman. A set of measures of centrality based on between- ness. Sociometry, 40(1):35–41, 1977

    work page 1977

Show all 52 references
  1. [8]

    S3: Social-network simulation system with large language model-empowered agents

    Chen Gao, Xiaochong Lan, Zhihong Lu, Jinzhu Mao, Jinghua Piao, Huandong Wang, Depeng Jin, and Yong Li. S3: Social-network simulation system with large language model-empowered agents. arXiv preprint arXiv:2307.14984, 2023

    2023 arXiv

  2. [9]

    Large language model based multi-agents: A survey of progress and challenges

    Taicheng Guo, Xiuying Chen, Yaqi Wang, Ruidi Chang, Shichao Pei, Nitesh V Chawla, Olaf Wiest, and Xiangliang Zhang. Large language model based multi-agents: A survey of progress and challenges. arXiv preprint arXiv:2402.01680, 2024

    2024 arXiv

  3. [10]

    The movielens datasets: History and context

    F Maxwell Harper and Joseph A Konstan. The movielens datasets: History and context. Acm transactions on interactive intelligent systems (tiis), 5(4):1–19, 2015

    2015

  4. [11]

    Lightgcn: Simplifying and powering graph convolution network for recommendation

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. Lightgcn: Simplifying and powering graph convolution network for recommendation. In Proceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, pages...

    2020

  5. [12]

    An llm-enhanced agent-based simulation tool for information propagation

    Yuxuan Hu, Gemju Sherpa, Lan Zhang, Weihua Li, Quan Bai, Yijun Wang, and Xiaodan Wang. An llm-enhanced agent-based simulation tool for information propagation. In Proceedings of the Thirty-Third International Joint Conference on Artificial Intelligence , IJCAI ’24, 2024

    2024

  6. [13]

    Recsim: A con- figurable simulation platform for recommender systems

    Eugene Ie, Chih-Wei Hsu, Martin Mladenov, Vihan Jain, Sanmit Narvekar, Jing Wang, Rui Wu, and Craig Boutilier. Recsim: A con- figurable simulation platform for recommender systems. In Proceedings of the 13th ACM Conference on Recommender Systems , pages 228–236. ACM, 2019

    2019

  7. [14]

    Positive affect, cognitive processes, and social behavior

    Alice M Isen. Positive affect, cognitive processes, and social behavior. In Advances in experimental social psychology , volume 20, pages 203–

  8. [15]

    Challenging low homophily in social recommendation

    Wei Jiang, Xinyi Gao, Guandong Xu, Tong Chen, and Hongzhi Yin. Challenging low homophily in social recommendation. In Proceedings of the ACM Web Conference 2024 , pages 3476–3484, 2024

    2024

  9. [16]

    Self-attentive sequential rec- ommendation

    Wang-Cheng Kang and Julian McAuley. Self-attentive sequential rec- ommendation. In 2018 IEEE international conference on data mining (ICDM), pages 197–206. IEEE, 2018

    2018

  10. [17]

    The foreign- language effect: Thinking in a foreign tongue reduces decision biases

    Boaz Keysar, Sayuri L Hayakawa, and Sun Gyu An. The foreign- language effect: Thinking in a foreign tongue reduces decision biases. Psychological science, 23(6):661–668, 2012

    2012

  11. [18]

    Matrix factorization techniques for recommender systems

    Yehuda Koren, Robert Bell, and Chris V olinsky. Matrix factorization techniques for recommender systems. Computer, 42(8):30–37, 2009

    2009

  12. [19]

    Private traits and attributes are predictable from digital records of human behavior

    Michal Kosinski, David Stillwell, and Thore Graepel. Private traits and attributes are predictable from digital records of human behavior. Proceedings of the National Academy of Sciences , 110(15):5802–5805, 2013

    2013

  13. [20]

    Learning to discover social circles in ego networks

    Jure Leskovec and Julian Mcauley. Learning to discover social circles in ego networks. Advances in neural information processing systems , 25, 2012

    2012

  14. [21]

    Variational autoencoders for collaborative filtering

    Dawen Liang, Rahul G Krishnan, Matthew D Hoffman, and Tony Jebara. Variational autoencoders for collaborative filtering. In Proceedings of the 2018 world wide web conference , pages 689–698, 2018

    2018

  15. [22]

    Hao Ma, Dengyong Zhou, Chao Liu, and Michael R. Lyu. Learning to recommend with social trust ensemble. In Proceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval, pages 203–210, 2009

    2009

  16. [23]

    Image-based recommendations on styles and substitutes

    Julian McAuley, Christopher Targett, Qinfeng Shi, and Anton Van Den Hengel. Image-based recommendations on styles and substitutes. In Proceedings of the 38th international ACM SIGIR conference on research and development in information retrieval , pages 43–52, 2015

    2015

  17. [24]

    Birds of a feather: Homophily in social networks

    Miller McPherson, Lynn Smith-Lovin, and James M Cook. Birds of a feather: Homophily in social networks. Annual review of sociology , 27(1):415–444, 2001

    2001

  18. [25]

    Recsim ng: Toward principled uncertainty modeling for recommender ecosystems

    Martin Mladenov, Craig Boutilier, and Eugene Ie. Recsim ng: Toward principled uncertainty modeling for recommender ecosystems. In Advances in Neural Information Processing Systems , volume 34, pages 14985–14997, 2021

    2021

  19. [26]

    Personality and domain-specific risk taking

    Nigel Nicholson, Emma Soane, Mark Fenton-O’Creevy, and Paul Will- man. Personality and domain-specific risk taking. Journal of Risk Research, 8(2):157–176, 2005

    2005

  20. [27]

    The pagerank citation ranking: Bringing order to the web

    Lawrence Page, Sergey Brin, Rajeev Motwani, and Terry Winograd. The pagerank citation ranking: Bringing order to the web. Technical report, Stanford InfoLab, 1999

    1999

  21. [28]

    Generative agents: Interactive simulacra of human behavior

    Joon Sung Park, Joseph C O’Brien, Carrie J Cai, Meredith Ringel Morris, Percy Liang, and Michael S Bernstein. Generative agents: Interactive simulacra of human behavior. In Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems , pages 1–18. ACM, 2023

    2023

  22. [29]

    Kullback-leibler divergence estimation of continu- ous distributions

    Fernando Pérez-Cruz. Kullback-leibler divergence estimation of continu- ous distributions. In 2008 IEEE international symposium on information theory, pages 1666–1670. IEEE, 2008

    2008

  23. [30]

    Agentsociety: Large-scale simulation of llm-driven generative agents advances understanding of human behaviors and society

    Jinghua Piao, Yuwei Yan, Jun Zhang, Nian Li, Junbo Yan, Xiaochong Lan, Zhihong Lu, Zhiheng Zheng, Jing Yi Wang, Di Zhou, et al. Agentsociety: Large-scale simulation of llm-driven generative agents advances understanding of human behaviors and society. arXiv preprint arXiv:2502...

    2025 arXiv

  24. [31]

    The earth mover’s distance as a metric for image retrieval

    Yossi Rubner, Carlo Tomasi, and Leonidas J Guibas. The earth mover’s distance as a metric for image retrieval. International journal of computer vision, 40:99–121, 2000

    2000

  25. [32]

    A circumplex model of affect

    James A Russell. A circumplex model of affect. Journal of personality and social psychology , 39(6):1161, 1980

    1980

  26. [33]

    Andrew Schwartz, Johannes C

    H. Andrew Schwartz, Johannes C. Eichstaedt, Margaret L. Kern, et al. Personality, gender, and age in the language of social media: The open- vocabulary approach. PLOS ONE, 8(9):e73791, 2013

    2013

  27. [34]

    Enhancing trust in llm- based ai automation agents: New considerations and future challenges

    Sivan Schwartz, Avi Yaeli, and Segev Shlomov. Enhancing trust in llm- based ai automation agents: New considerations and future challenges. arXiv preprint arXiv:2308.05391 , 2023

    2023 arXiv

  28. [35]

    A survey of graph neural networks for social recommender systems

    Kartik Sharma, Yeon-Chang Lee, Sivagami Nambi, Aditya Salian, Shlok Shah, Sang-Wook Kim, and Srijan Kumar. A survey of graph neural networks for social recommender systems. ACM Computing Surveys , 56(10):1–34, 2024

    2024

  29. [36]

    A first principles approach to trust-based recommendation systems

    Paras Stefanopoulos, Ahad N Zehmakan, and Sourin Chatterjee. A first principles approach to trust-based recommendation systems. arXiv preprint arXiv:2407.00062, 2024

    2024 arXiv

  30. [37]

    The clarion cognitive architecture: Extending cognitive modeling to social simulation

    Ron Sun. The clarion cognitive architecture: Extending cognitive modeling to social simulation. Cognition and multi-agent interaction , pages 79–99, 2006

    2006

  31. [38]

    Gensim: A general social simulation platform with large language model based agents

    Jiakai Tang, Heyang Gao, Xuchen Pan, Lei Wang, Haoran Tan, Dawei Gao, Yushuo Chen, Xu Chen, Yankai Lin, Yaliang Li, et al. Gensim: A general social simulation platform with large language model based agents. arXiv preprint arXiv:2410.04360 , 2024

    arXiv

  32. [39]

    Social recommendation: a review

    Jiliang Tang, Xia Hu, and Huan Liu. Social recommendation: a review. Social Network Analysis and Mining , 3(4):1113–1133, 2013

    2013

  33. [40]

    Learning to rank by optimizing ndcg measure

    Hamed Valizadegan, Rong Jin, Ruofei Zhang, and Jianchang Mao. Learning to rank by optimizing ndcg measure. Advances in neural information processing systems , 22, 2009

    2009

  34. [41]

    V oyager: An open- ended embodied agent with large language models

    Guanzhi Wang, Yuqi Xie, Yunfan Jiang, Ajay Mandlekar, Chaowei Xiao, Yuke Zhu, Linxi Fan, and Anima Anandkumar. V oyager: An open- ended embodied agent with large language models. arXiv preprint arXiv:2305.16291, 2023

    2023 arXiv

  35. [42]

    User behavior simulation with large language model based agents

    Lei Wang, Jingsen Zhang, Hao Yang, Zhiyuan Chen, Jiakai Tang, Zeyu Zhang, Xu Chen, Yankai Lin, Ruihua Song, Wayne Xin Zhao, et al. User behavior simulation with large language model based agents. arXiv preprint arXiv:2306.02552, 2023

    2023 arXiv

  36. [43]

    Impact of personal traits and social networks on online knowledge sharing: A case study of facebook users

    Yu Wang, Youcheng Wang, and Wen Wang. Impact of personal traits and social networks on online knowledge sharing: A case study of facebook users. Computers in Human Behavior , 34:345–354, 2014

    2014

  37. [44]

    Can large language model agents simulate human trust behavior? In The Thirty- eighth Annual Conference on Neural Information Processing Systems , 2024

    Chengxing Xie, Canyu Chen, Feiran Jia, Ziyu Ye, Shiyang Lai, Kai Shu, Jindong Gu, Adel Bibi, Ziniu Hu, David Jurgens, et al. Can large language model agents simulate human trust behavior? In The Thirty- eighth Annual Conference on Neural Information Processing Systems , 2024

    2024

  38. [46]

    Behavior alignment: A new perspective of evaluating llm-based conversational recommender sys- tems

    Dayu Yang, Fumian Chen, and Hui Fang. Behavior alignment: A new perspective of evaluating llm-based conversational recommender sys- tems. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages 2286–

  39. [47]

    Invariance matters: Empowering social recommendation via graph invariant learning

    Yonghui Yang, Le Wu, Yuxin Liao, Zhuangzhuang He, Pengyang Shao, Richang Hong, and Meng Wang. Invariance matters: Empowering social recommendation via graph invariant learning. arXiv preprint arXiv:2504.10432, 2025

    2025 arXiv

  40. [48]

    Oasis: Open agents social interaction simulations on one million agents

    Ziyi Yang, Zaibin Zhang, Zirui Zheng, Yuxian Jiang, Ziyue Gan, Zhiyu Wang, Zijian Ling, Jinsong Chen, Martz Ma, Bowen Dong, et al. Oasis: Open agents social interaction simulations on one million agents. arXiv preprint arXiv:2411.11581, 2024

    2024 arXiv

  41. [49]

    On generative agents in recommendation

    An Zhang, Yuxin Chen, Leheng Sheng, Xiang Wang, and Tat-Seng Chua. On generative agents in recommendation. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages 1234–1243. ACM, 2023

    2023

  42. [50]

    Llm- aidsim: Llm-enhanced agent-based influence diffusion simulation in social networks

    Lan Zhang, Yuxuan Hu, Weihua Li, Quan Bai, and Parma Nand. Llm- aidsim: Llm-enhanced agent-based influence diffusion simulation in social networks. Systems, 13(1):29, 2025

    2025

  43. [51]

    Impact of introversion- extraversion personality traits on knowledge sharing: Evidence from virtual communities

    Min Zhang, Ji Liang, and Xiaofang Wu. Impact of introversion- extraversion personality traits on knowledge sharing: Evidence from virtual communities. Sustainability, 15(1):417, 2023

    2023

  44. [52]

    Kuaisim: A comprehensive simulator for recommender systems

    Kesen Zhao, Shuchang Liu, Qingpeng Cai, Xiangyu Zhao, Ziru Liu, Dong Zheng, Peng Jiang, and Kun Gai. Kuaisim: A comprehensive simulator for recommender systems. In Proceedings of the 17th ACM Conference on Recommender Systems , pages 456–466. ACM, 2023

    2023

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.