Pith. sign in

REVIEW 3 major objections 5 minor 45 references

Personalized Communication Skills for Agentic Recommender Systems

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read AgentCom claims that adding personalized why-what-how-who communication skills to UserAgents improves recommendation accuracy across traditional, social, and agentic recommenders.

desk verdict AgentCom is a coherent communication-skill layer for agentic recommenders, but its headline 'consistently improves' rests entirely on single-run Hit@1 numbers with no error bars, so treat the gains as unverified until repeated runs and significance tests appear. read the letter →

arxiv 2608.08417 v1 pith:EV2FTYDI submitted 2026-08-09 cs.IR

Zongwei Wang , Min Gao , Guangyu Hu , Xinyi Gao , Junliang Yu This is my paper · ORCID
classification cs.IR
keywords agenticrecommendersystemsuseragentsadvisorcommunicationskillswhy-what-how-whoskillbankpersonalizedroutingfailure-drivenevolutionlargelanguagemodels
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

This paper argues that UserAgents—large-language-model simulations of a person used to test candidate items before a recommendation is delivered—choose better when they can consult other users' agents as advisors, and that the consultation has to be matched to the user's current decision problem instead of following a fixed script. To make that work, the authors build AgentCom, a layer that sits on top of any recommender and organizes advisor communication as reusable skills along four questions: why advice is needed, what information to request, how advisors should interact, and who should advise. The paper reports consistent Hit@1 improvements when AgentCom is added to all five backbones tested on three datasets, including raising the sequential model SASRec's LastFM accuracy from 0.0155 to 0.2019 and improving the strongest agentic baseline MemRec. If the measurements hold, cross-user agent communication becomes a general plug-in that can upgrade existing recommenders without retraining them.

What carries the argument

The load-bearing object is the why–what–how–who communication skill bank: a hierarchy of reusable nodes, each holding an applicability condition, an executable instruction, and an output contract. A complete communication path is one selection from each layer, and an LLM-based router ranks candidates for the next layer conditioned on the selection made in the layer before it. This conditional ranking is what makes the shared bank personalized at use time. Around the bank sit two mechanisms: personalized skill routing, which builds each user's path by traversing the conditional ranked lists, and failure-driven skill evolution, which diagnoses failed decisions by layer and responds with skill refinement, skill generation, or rerouting. The whole design translates the abstract idea that different decision states need different advice into an executable loop that can be trained and then fixed for inference.

What would settle it

Re-run the Table 2 protocol with at least ten random seeds and report mean, standard deviation, and confidence intervals; if the AgentCom-over-baseline gaps for SASRec and GBSR overlap zero, or if removing the guaranteed ground-truth item from the 20-candidate set eliminates the advantage, the central claim is not supported.

Watch

Extended reading notes

Core claim

AgentCom's central claim is that UserAgents make more accurate final selections when they are allowed to gather evidence from other users' agents, and that this communication must be assembled from reusable skills rather than run as a generic dialogue. The framework decomposes every communication act along four questions: why the target agent needs external advice, what information should be requested, how the advisors should interact, and who the suitable advisors are. Each skill is a node with an applicability condition, an executable instruction, and an output contract; a shared bank stores these nodes hierarchically, personalized routing picks a path for each user and context, and failure-driven evolution adds or refines skills when final choices miss the ground-truth item. The paper's evidence is a set of Hit@1 comparisons in which adding AgentCom improves every backbone on every dataset, including the strongest baseline MemRec.

Load-bearing premise

The load-bearing assumption is that the reported Hit@1 gains are caused by AgentCom's communication skills and not by chance variation in the language model's outputs or by the protocol that always inserts the correct item into the candidate set.

Editorial extensions

If this is right

  • Adding AgentCom to a non-agentic sequential recommender such as SASRec produces the largest relative gains, so even simple backbones can be upgraded by LLM-mediated advisor evidence without retraining.
  • Adding AgentCom to already-agentic recommenders such as MemRec still helps, so cross-user communication is complementary to interaction memory and internal feedback rather than a substitute for them.
  • Removing any single layer of the why–what–how–who structure lowers Hit@1, implying each layer answers a distinct question that generic dialogue would leave unresolved.
  • Disabling failure-driven evolution lowers performance, so a fixed skill bank is not enough; the framework's improvement depends on updating skills from unsuccessful cases.
  • More user–advisor interaction rounds help up to a point, while more advisor-only discussion consistently hurts, so the cost-quality trade-off has an identifiable optimum.

Reading between the lines

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

  • A testable extension the paper does not run is replacing natural-language advisor messages with compact preference vectors; if the gains survive, the value lies in the evidence content rather than in LLM deliberation, and the cost per user would drop sharply.
  • Because the protocol always places the ground-truth item in the 20-candidate set, the reported Hit@1 is an upper-bound-style accuracy; a harder deployment test would draw candidate sets without the known answer and measure how much of the gain remains.
  • The failure-driven evolution loop is stated for recommendation, but it is a generic recipe for any agentic decision process where a final choice can be checked against an outcome; transferring the why–what–how–who skill bank to other domains is a direct corollary.
  • The who layer currently mixes social links and embedding similarity; varying the advisor source alone would reveal whether the gains come from trusted relationships, from preference overlap, or from both.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes AgentCom, a framework that inserts a personalized communication-skill layer into agentic recommender systems. AgentCom maintains a shared why--what--how--who skill bank, routes each user and recommendation context through a personalized communication path, and evolves the bank from training-time failures via skill refinement, skill generation, and routing correction. Experiments on LastFM, Epinions, and LibraryThing compare five backbones (SASRec, GBSR, AFL, iAgent, MemRec) with and without AgentCom, reporting Hit@1 in Table 2, plus ablations, hyperparameter analysis, a case study, and an LLM cost analysis. The central claim is that AgentCom consistently improves recommendation performance across traditional, social, and agentic recommenders.

Significance. If the empirical claim is supported, the paper makes a useful conceptual contribution: framing advisor-agent communication as reusable, hierarchically organized skills with a why--what--how--who decomposition, and complementing it with failure-driven skill evolution. This is a plausible and timely direction for agentic recommender systems, and the shared-bank, personalized-routing design is a sensible way to make communication adaptive without per-user prompt engineering. The paper also ships a complete algorithmic description and cost analysis, which strengthens reproducibility. However, the main claim rests on single-run Hit@1 numbers without uncertainty quantification and on comparisons that lack a non-communication LLM reranking baseline, so the current evidence does not yet establish that the observed improvements are due to the proposed skill structure.

major comments (3)
  1. [Section 4.2, Table 2] The central claim that AgentCom "consistently improves" performance across all backbones and datasets is supported only by single-run Hit@1 values. No standard deviations, confidence intervals, significance tests, random seeds, or sampling temperatures are reported, despite the pipeline being composed of stochastic LLM components (UserAgents, advisor agents, router, and failure analyzer). Several differences are small in absolute terms, for example MemRec on LastFM improves from 0.2648 to 0.2696 (+0.0048) and on Epinions from 0.2427 to 0.2485 (+0.0058); these could easily be within run-to-run noise. To support the headline claim, the authors should run multiple independent seeds, report means and spreads, and perform paired significance tests such as McNemar's test or a bootstrap over users.
  2. [Section 4.1.4 and ablation in Section 4.3] There is no baseline that reranks the same 20-item candidate set with the same underlying LLM but without any communication-skill structure. The "w/o all" ablation replaces the structured skill with generic dialogue, but it is not described precisely: it must be stated whether generic dialogue still includes advisor agents and, if so, how advisors are selected. Without a pure LLM-reranking baseline, the measured gains could be attributed to the LLM's ability to evaluate candidates rather than to the why--what--how--who skill decomposition. The authors should add a no-communication LLM reranker and a generic single-advisor dialogue variant to isolate the contribution of the skill structure.
  3. [Section 4.4, Figure 4] The hyperparameter conclusions about Kmax and G are based on single-run Hit@1 differences as small as 0.0039 and 0.0048, with no error bars or significance tests. The claim that increasing G reduces performance and that Kmax = 3 is the right operating point should be supported by repeated runs or confidence intervals, especially because the differences are comparable in magnitude to the small gains in Table 2.
minor comments (5)
  1. [Section 4.1.4] The evaluation protocol always inserts the ground-truth item into the 20-item candidate set. This is fair for within-backbone comparisons because both the original and AgentCom-enhanced variants operate on the same candidate set, but it means the absolute Hit@1 values in Table 2 are not directly comparable to standard top-k metrics in the literature; the authors should state this limitation explicitly.
  2. [Table 2 caption] The caption says that AgentCom-enhanced variants are "highlighted in blue, with their results shown in bold," but the table as presented has no visible blue highlighting or bold formatting; this is a typesetting issue that should be corrected.
  3. [Section 4.3, Figure 3] The ablation figure would be more informative if accompanied by a table of exact Hit@1 values, since the current text reports only qualitative decreases and the figure is hard to read at low resolution.
  4. [Section 3.2.1, Eq. (12)] The RankLLM operation is central to routing, but the paper does not specify the prompt template, the number of nodes given to the router, or the temperature used; adding these details would substantially improve reproducibility.
  5. [Section 2.2] The notation K_{u,t} is introduced as the number of communication rounds and later reused in Algorithm 1 and the hyperparameter section as Kmax; the relation between K_{u,t} and Kmax should be stated explicitly in Section 2.2.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: AgentCom is an empirical framework evaluated against external backbones; no prediction reduces to its inputs by construction.

full rationale

AgentCom is an empirical framework rather than a derivational one: no equation in the paper defines a predicted quantity in terms of the fitted skill bank or the ground-truth labels. The why–what–how–who decomposition is explicitly presented as an organization inspired by Lasswell's communication model (cited as [13]), and the initial skill bank is theory-guided (Appendix A), with the actual routing and evolution performed by LLM calls on training interactions (Eqs. 12–25). Failure-driven evolution uses the ground-truth item only to identify training failures (Eq. 15) and to generate or refine skills during training; the resulting bank is fixed for inference (Algorithm 1: 'After R evolution rounds, the resulting skill bank is fixed for inference'), which is standard supervised practice. The headline improvements in Table 2 are measured against external backbones on test instances under a shared candidate set; although one could question the fairness of always inserting the ground-truth item into the candidate set or the absence of error bars, those are experimental-design concerns rather than circular derivation. No self-citation is load-bearing: the only author self-citation [31] appears in Related Work for the one-shot prediction paradigm and does not support any of AgentCom's central claims. Therefore, no step in the paper reduces to its own inputs by construction, and the central claim remains an empirical result rather than a tautology.

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

AgentCom introduces no new physical or mathematical entities. The communication skills are software structures (applicability condition, executable instruction, output contract) rather than invented postulates with independent evidence requirements. The free parameters are hyperparameters and evaluation-design choices. The axioms are largely domain assumptions about LLM reliability and the validity of the evaluation protocol.

free parameters (4)
  • Kmax = 3
    Maximum number of user-advisor communication rounds, selected via the hyperparameter sweep in Section 4.4; increasing beyond 3 gives diminishing returns and higher LLM cost.
  • G = 1
    Number of internal advisor discussion rounds, selected via the hyperparameter sweep in Section 4.4; increasing G reduces Hit@1 on all datasets.
  • Candidate set size = 20
    Evaluation design choice in Section 4.1.4: the ground-truth item plus the 19 highest-ranked non-ground-truth items. This conditions Hit@1 on the ground truth being present in the candidate set.
  • Number of evolution rounds R
    The paper never specifies R, the number of failure-driven evolution rounds, nor the resulting bank size; this missing design parameter affects the final skill bank used at inference.
assumptions (4)
  • domain assumption Advisor agents with diverse histories provide complementary evidence that improves target-user judgment in this simulated setting.
    Borrowed from human advice-taking literature [36,37], asserted without direct validation in Section 1 and used as the motivation for the whole framework.
  • ad hoc to paper The why-what-how-who decomposition is a sufficient and complete structure for agent communication.
    Introduced in Section 3.1; the paper gives no argument that other dimensions (e.g., when to stop, how to weight advisors) are unnecessary.
  • domain assumption The LLM (deepseek-4-flash) reliably performs routing, failure diagnosis, skill refinement, and skill generation as intended.
    All core operations are delegated to an LLM (Sections 3.2.1 and 3.2.2); no mechanism is provided to verify the correctness of routing or generated skills.
  • domain assumption Hit@1 on a candidate set that always contains the ground-truth item is a valid measure of recommendation performance for this claim.
    The shared-candidate protocol in Section 4.1.4 inserts the ground-truth item into every candidate set, so Hit@1 is conditional on successful retrieval; this is not discussed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Personalized Communication Skills for Agentic Recommender Systems." pith.science (2026). https://pith.science/paper/EV2FTYDI

@misc{pith2026260808417,
  author       = {Pith},
  title        = {Pith review of: Personalized Communication Skills for Agentic Recommender Systems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EV2FTYDI}},
  note         = {Machine review of arXiv:2608.08417}
}
read the original abstract

Agentic recommender systems increasingly employ large language model-based UserAgents to evaluate candidate items through simulated feedback before recommendations are delivered. However, existing UserAgents typically reason in isolation based on limited personal histories, which may lead to perspective narrowing: the agent evaluates candidates from a local and incomplete view, overlooks relevant preference facets, and consequently produces inaccurate judgments. A natural way to alleviate this problem is to introduce other users as advisor agents, whose diverse histories provide complementary evidence that helps the target user reconsider overlooked preference signals. Nevertheless, a generic user-advisor communication process is insufficient, as different user decision states require different forms of external advice. Based on this insight, we propose AgentCom, a personalized communication skill framework for agentic recommender systems. AgentCom organizes reusable communication skills into a shared why--what--how--who skill bank: why identifies the decision deficiency that necessitates communication, what specifies the information task, how determines the advisor interaction protocol, and who retrieves advisors capable of executing that protocol. To make the shared skill bank personalized at use time and adaptive over time, AgentCom introduces two complementary mechanisms: personalized skill routing and failure-driven skill evolution. Personalized skill routing constructs a communication path by sequentially selecting suitable skills for each user and recommendation context. Failure-driven skill evolution learns from unsuccessful communication cases and enriches the shared bank with reusable skills that address previously uncovered communication needs. Experiments show that AgentCom consistently improves recommendation performance across traditional, social, and agentic recommenders.

Figures

Figures reproduced from arXiv: 2608.08417 by the authors.

Figure 1
Figure 1. Comparison between independent judgment and [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Overview of AgentCom. The target UserAgent first evaluates recommender-generated candidates. AgentCom selects [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Ablation results of AgentCom on three datasets. [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Effect of the maximum number of user–advisor [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

45 extracted references · 19 canonical work pages

  1. [1]

    Dirk Bollen, Bart P Knijnenburg, Martijn C Willemsen, and Mark Graus. 2010. Understanding choice overload in recommender systems. InProceedings of the fourth ACM conference on Recommender systems. 63–70

    work page 2010

  2. [2]

    Chenwei Cai, Ruining He, and Julian McAuley. 2017. SPMC: Socially-aware personalized Markov chains for sparse sequential recommendation.arXiv preprint arXiv:1708.04497(2017)

    arXiv 2017

  3. [3]

    Shihao Cai, Jizhi Zhang, Keqin Bao, Chongming Gao, Qifan Wang, Fuli Feng, and Xiangnan He. 2025. Agentic feedback loop modeling improves recommendation and user simulation. InProceedings of the 48th International ACM SIGIR conference on Research and Development in Information Retrieval. 2235–2244

    2025

  4. [4]

    Iván Cantador, Peter Brusilovsky, and Tsvi Kuflik. 2011. Second workshop on information heterogeneity and fusion in recommender systems (HetRec2011). In Proceedings of the fifth ACM conference on Recommender systems. 387–388

    2011

  5. [5]

    Weixin Chen, Yuhan Zhao, Jingyuan Huang, Zihe Ye, Mingxuan Ju, Tong Zhao, Neil Shah, Li Chen, and Yongfeng Zhang. 2026. Memrec: Collaborative memory- augmented agentic recommender system. InProceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 44515–44544

    work page 2026

  6. [6]

    Yilun Du, Shuang Li, Antonio Torralba, Joshua B Tenenbaum, and Igor Mordatch

  7. [7]

    Mark S Granovetter. 1973. The strength of weak ties.American journal of sociology 78, 6 (1973), 1360–1380

    1973

  8. [8]

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Peiyi Wang, Qihao Zhu, Runxin Xu, Ruoyu Zhang, Shirong Ma, Xiao Bi, et al . 2025. DeepSeek-R1 in- centivizes reasoning in LLMs through reinforcement learning.Nature645, 8081 (2025), 633–638

    2025

Show all 45 references
  1. [9]

    John R Hauser and Birger Wernerfelt. 1990. An evaluation cost model of consid- eration sets.Journal of consumer research16, 4 (1990), 393–408

    1990

  2. [10]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yongdong Zhang, and Meng Wang. 2020. Lightgcn: Simplifying and powering graph convolution network for recommendation. InProceedings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval. 639–648

    2020

  3. [11]

    Wang-Cheng Kang and Julian McAuley. 2018. Self-attentive sequential recom- mendation. In2018 IEEE international conference on data mining (ICDM). IEEE, 197–206

    2018

  4. [12]

    Richard P Larrick. 2004. Debiasing.Blackwell handbook of judgment and decision making(2004), 316–338

    2004

  5. [13]

    Harold D Lasswell. 1948. The structure and function of communication in society. The communication of ideas37, 1 (1948), 136–139

    1948

  6. [14]

    Kyle Lewis. 2004. Knowledge and performance in knowledge-worker teams: A longitudinal study of transactive memory systems.Management science50, 11 (2004), 1519–1533

    2004

  7. [15]

    Bingqian Li, Xiaolei Wang, Junyi Li, Weitao Li, Long Zhang, Sheng Chen, Wayne Xin Zhao, and Ji-Rong Wen. 2026. RecNet: Self-Evolving Preference Propagation for Agentic Recommender Systems.arXiv preprint arXiv:2601.21609 (2026)

    2026

  8. [16]

    Tian Liang, Zhiwei He, Wenxiang Jiao, Xing Wang, Yan Wang, Rui Wang, Yujiu Yang, Shuming Shi, and Zhaopeng Tu. 2024. Encouraging divergent thinking in large language models through multi-agent debate. InProceedings of the 2024 conference on empirical methods in natural languag...

    2024

  9. [17]

    Jiahao Liu, Shengkang Gu, Dongsheng Li, Guangping Zhang, Mingzhe Han, Hansu Gu, Peng Zhang, Tun Lu, Li Shang, and Ning Gu. 2025. AgentCF++: Memory-enhanced LLM-based Agents for Popularity-aware Cross-domain Rec- ommendations. InProceedings of the 48th International ACM SIGIR C...

    2025

  10. [18]

    George Loewenstein. 1994. The psychology of curiosity: A review and reinter- pretation.Psychological bulletin116, 1 (1994), 75

    1994

  11. [19]

    Hao Ma, Irwin King, and Michael R Lyu. 2009. Learning to recommend with social trust ensemble. InProceedings of the 32nd international ACM SIGIR conference on Research and development in information retrieval. 203–210

    2009

  12. [20]

    Jessica R Mesmer-Magnus and Leslie A DeChurch. 2009. Information sharing and team performance: a meta-analysis.Journal of applied psychology94, 2 (2009), 535

    2009

  13. [21]

    2017.An analysis of Leon Festinger’s a theory of cognitive dissonance

    Camille Morvan and Alexander O’Connor. 2017.An analysis of Leon Festinger’s a theory of cognitive dissonance. Macat Library

    2017

  14. [22]

    Charlan Nemeth and John Rogers. 1996. Dissent and the search for information. British Journal of Social Psychology35, 1 (1996), 67–76

    1996

  15. [23]

    1993.The adaptive decision maker

    John W Payne, James R Bettman, and Eric J Johnson. 1993.The adaptive decision maker. Cambridge university press

    1993

  16. [24]

    Paul Resnick, Neophytos Iacovou, Mitesh Suchak, Peter Bergstrom, and John Riedl. 1994. Grouplens: An open architecture for collaborative filtering of netnews. InProceedings of the 1994 ACM conference on Computer supported cooperative work. 175–186

    1994

  17. [25]

    Andrew I Schein, Alexandrin Popescul, Lyle H Ungar, and David M Pennock

  18. [26]

    Janet A Sniezek and Lyn M Van Swol. 2001. Trust, confidence, and expertise in a judge-advisor system.Organizational behavior and human decision processes84, 2 (2001), 288–307

    2001

  19. [27]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang

  20. [28]

    Amos Tversky and Eldar Shafir. 1992. Choice under conflict: The dynamics of deferred decision.Psychological science3, 6 (1992), 358–361

    1992

  21. [29]

    Lei Wang, Jingsen Zhang, Hao Yang, Zhi-Yuan Chen, Jiakai Tang, Zeyu Zhang, Xu Chen, Yankai Lin, Hao Sun, Ruihua Song, et al. 2025. User behavior simulation with large language model-based agents.ACM Transactions on Information Systems43, 2 (2025), 1–37

    2025

  22. [30]

    Yancheng Wang, Ziyan Jiang, Zheng Chen, Fan Yang, Yingxue Zhou, Eunah Cho, Xing Fan, Yanbin Lu, Xiaojiang Huang, and Yingzhen Yang. 2024. Recmind: Large language model powered agent for recommendation. InFindings of the Association for Computational Linguistics: NAACL 2024. 4351–4364

    2024

  23. [31]

    Zongwei Wang, Min Gao, Wentao Li, Junliang Yu, Linxin Guo, and Hongzhi Yin

  24. [32]

    Stacy L Wood and C Page Moreau. 2006. From fear to loathing? How emotion influences the evaluation and early use of innovations.Journal of Marketing70, 3 (2006), 44–57

    2006

  25. [33]

    Yu Xia, Sungchul Kim, Tong Yu, Ryan A Rossi, and Julian McAuley. 2025. Multi- Agent Collaborative Filtering: Orchestrating Users and Items for Agentic Recom- mendations.arXiv preprint arXiv:2511.18413(2025)

    2025

  26. [34]

    Wujiang Xu, Yunxiao Shi, Zujie Liang, Xuying Ning, Kai Mei, Kun Wang, Xi Zhu, Min Xu, and Yongfeng Zhang. 2025. iagent: Llm agent as a shield between user and recommender systems. InFindings of the Association for Computational Linguistics: ACL 2025. 18056–18084

    2025

  27. [35]

    Yonghui Yang, Le Wu, Zihan Wang, Zhuangzhuang He, Richang Hong, and Meng Wang. 2024. Graph bottlenecked social recommendation. InProceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 3853–3862

    2024

  28. [36]

    Ilan Yaniv. 2004. Receiving other people’s advice: Influence and benefit.Organi- zational behavior and human decision processes93, 1 (2004), 1–13

    2004

  29. [37]

    Ilan Yaniv and Maxim Milyavsky. 2007. Using advice from multiple sources to revise and improve judgments.Organizational Behavior and Human Decision Processes103, 1 (2007), 104–120

    2007

  30. [38]

    An Zhang, Yuxin Chen, Leheng Sheng, Xiang Wang, and Tat-Seng Chua. 2024. On generative agents in recommendation. InProceedings of the 47th international Conference acronym ’XX, June 03–05, 2018, Woodstock, NY Trovato et al. ACM SIGIR conference on research and development in I...

    2024

  31. [39]

    Junjie Zhang, Yupeng Hou, Ruobing Xie, Wenqi Sun, Julian McAuley, Wayne Xin Zhao, Leyu Lin, and Ji-Rong Wen. 2024. Agentcf: Collaborative learning with autonomous language agents for recommender systems. InProceedings of the ACM Web Conference 2024. 3679–3689

    2024

  32. [40]

    Tong Zhao, Julian McAuley, and Irwin King. 2015. Improving latent factor models via personalized feature projection for one class recommendation. InProceed- ings of the 24th ACM international on conference on information and knowledge management. 821–830

    2015

  33. [41]

    Yuyue Zhao, Jiancan Wu, Xiang Wang, Wei Tang, Dingxian Wang, and Maarten De Rijke. 2024. Let me do it for you: Towards llm empowered recommendation via tool learning. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrie...

    2024

  34. [2002]

    InProceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval

    Methods and metrics for cold-start recommendations. InProceedings of the 25th annual international ACM SIGIR conference on Research and development in information retrieval. 253–260

  35. [2019]

    InProceedings of the 28th ACM international conference on information and knowledge management

    BERT4Rec: Sequential recommendation with bidirectional encoder rep- resentations from transformer. InProceedings of the 28th ACM international conference on information and knowledge management. 1441–1450

  36. [2023]

    InProceedings of the 29th ACM SIGKDD conference on knowledge discovery and data mining

    Efficient bi-level optimization for recommendation denoising. InProceedings of the 29th ACM SIGKDD conference on knowledge discovery and data mining. 2502–2511

  37. [2024]

    InProceedings of the 41st International Conference on Machine Learning

    Improving factuality and reasoning in language models through multiagent debate. InProceedings of the 41st International Conference on Machine Learning. 11733–11763

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.