Pith. sign in

REVIEW 2 major objections 6 minor 184 references

Agentic recommender systems are best understood through a two-axis taxonomy of agent role and autonomy level, which unifies the field and points the way to more goal-directed recommendation agents.

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 ↗

T0 review · grok-4.5

2026-07-11 19:12 UTC pith:VBOCWYVG

load-bearing objection Solid field-organizing survey: role × LoA taxonomy plus evaluation matrix give the exploding agentic-RecSys literature a usable shared vocabulary without overclaiming novelty. the 2 major comments →

arxiv 2607.04433 v1 pith:VBOCWYVG submitted 2026-07-05 cs.IR cs.CL

Autonomous Information Seeking: A Roadmap for Agentic Recommender Systems

Xinyu Lin , Yashar Deldjoo , Sunhao Dai , Honghui Bao , Xiaopeng Ye , Fatemeh Nazary , Wenjie Wang , Tommaso Di Noia
show 2 more authors
Jun Xu Tat-Seng Chua
This is my paper
classification cs.IR cs.CL
keywords Agentic Recommender SystemLevel of AutonomyLLM AgentAgent-assisted RecommenderAgent as RecommenderAgent as User SimulatorTool use and memoryEvaluation of agentic systems
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.

This survey argues that large language model agents are turning recommender systems from static rankers into systems that reason, plan, and act. It organizes the literature by three roles an agent can play—assisting a classical pipeline, fully replacing the recommender, or simulating users and environments—and by a level-of-autonomy ladder that tracks rising proactivity, context awareness, interaction flexibility, and adaptivity. The same framework is then used to compare how agents handle profiles, memory, tools, workflows, and optimization, and to diagnose why current evaluation metrics miss reasoning quality, trajectory behavior, and simulator fidelity. The authors claim this dual taxonomy supplies a shared map of progress and open problems so the community can build more autonomous, reliable, and human-aligned recommenders.

Core claim

The paper claims that agentic recommender systems form a coherent research area that is usefully partitioned by macro role (agent-assisted, agent-as-recommender, agent-as-user-simulator) crossed with Levels of Autonomy L2–L5, and that this partition plus the accompanying analysis of components, evaluation gaps, and open challenges establishes a unified foundation for the field.

What carries the argument

The two-axis taxonomy: macro role of the agent inside the recommendation loop, crossed with a Level-of-Autonomy spectrum (L2 retrieval-grounded, L3 tool-orchestrating, L4 single-agent planning, L5 multi-agent orchestration) defined by proactivity, context awareness, interaction flexibility, and adaptivity.

Load-bearing premise

That the four autonomy dimensions and the discrete L2–L5 cut-points are the most natural and stable way to organize the literature, rather than task type, domain, or architecture.

What would settle it

A systematic re-coding of the same paper corpus under an alternative primary axis (for example pure task taxonomy or architecture) that yields cleaner clusters, fewer borderline cases, and stronger predictive power for evaluation needs and open challenges than the role-by-LoA matrix.

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

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

2 major / 6 minor

Summary. This survey argues that LLM-based agentic recommender systems form a coherent research area that is usefully organized by a dual taxonomy: three macro roles (agent-assisted recommendation, agent-as-recommender, agent-as-user-simulator) crossed with a Level-of-Autonomy spectrum focused on L2–L5 (retrieval-grounded, tool-orchestrating, single-agent planning, multi-agent orchestration). Definitions 2.1–2.2 formalize LLM agents and ARS as interactive decision processes; §2.3 and Figures 3–4 map the literature onto roles and autonomy; §§3–5 review architectures and components (profile, memory, tools, workflow, optimization); §6 treats evaluation targets, protocols, and metrics, including trajectory-level and simulator-validity gaps; §7 lists open challenges in lifelong modeling, context, multimodality, controllability, trust, privacy, scalability, and efficiency. Search methodology (§1.4) and literature statistics (Figure 2) support the claim of rapid growth and a shift toward higher-autonomy, goal-driven systems.

Significance. The paper addresses a timely and fast-moving area in information retrieval. Its main contribution is shared vocabulary: the role×LoA frame, the component decomposition for single- and multi-agent recommenders, and a layered evaluation matrix (Table 6) that separates final ranking utility from process quality (tools, memory, planning, coordination, safety, cost). The systematic search description, explicit positioning against prior LLM/agent RS surveys (Table 2), and the evaluation checklist (Table 7) are practical strengths that can reduce fragmented terminology and under-specified claims in follow-on work. For a survey whose product is organization and gap identification rather than a new algorithm, residual subjectivity in binning systems is expected and does not erase the organizational value. If adopted, the taxonomy and evaluation stack would help the community compare agentic designs more fairly and avoid evaluating multi-step agents solely with static top-N metrics.

major comments (2)
  1. §2.3 and §§3–4: Several systems (e.g., InteRecAgent, AgentCF/AgentCF++, iAgent, AgentDR) are discussed under both agent-assisted and agent-as-recommender (and appear in multiple branches of Figure 3). The continuum framing of Figure 1 and the role illustrations in Figure 4 acknowledge overlap, but the survey does not state an explicit multi-label or primary-role rule. For the dual taxonomy to remain a stable organizing device, please add a short classification protocol (primary decision authority vs. optional multi-label) and apply it consistently when systems straddle augmentation and replacement.
  2. §1.4 and Figure 2: The quantitative claims (≈3× growth 2024→2025; Agent-as-Recommender dominance; L4/L5 share trends) rest on author annotation of role and LoA. The inclusion criteria and coding dimensions are stated, but inter-annotator agreement, borderline-case handling, and a public paper list or coding sheet are not. These statistics are secondary to the conceptual taxonomy, yet they are used to support “shift towards higher-autonomy” narratives. Either release the annotated corpus (or a stable appendix list) or qualify Figure 2 more carefully as illustrative rather than definitive trend evidence.
minor comments (6)
  1. §3 opening and §3.3: Incomplete or broken sentences remain (e.g., “leveraging agent as engineer to update the recommendation strategies?.” and “Rrepresentation planning”). Please proofread for truncated phrases and doubled letters.
  2. Table 3 and §3.2: Multimodal tools cite HuggingGPT and Toolformer as representative of tool-driven assistance in RS; these are general tool-use papers. Prefer RS-specific multimodal tool examples already listed elsewhere (e.g., VRAgent-R1, RAG-VisualRec) for consistency with the survey’s scope.
  3. §4.2.4 / Table 5: “Agen-tRecBench” and similar line-break artifacts appear in the workflow table; clean hyphenation and ensure every cited system has a matching bibliography entry.
  4. Figure 2 caption: The caption says “Left: distribution over different levels of autonomy. Right: distribution over different roles of agents,” but the panel labels in the text reverse this order relative to the figure description in §1.4. Align caption and body text.
  5. §6.1 and Table 6: The H0–H3 labels are introduced without a one-sentence definition of what H0–H3 stand for. A brief gloss would help readers who skip the surrounding prose.
  6. Throughout: Occasional citation-year mismatches and arXiv-style IDs in the main text are fine for a preprint, but for journal form normalize author–year citations and ensure every in-text key appears in the reference list.

Circularity Check

0 steps flagged

No circular derivation: survey taxonomy organizes third-party literature without fitted predictions or self-referential equations.

full rationale

This is a CS.IR survey that proposes a dual taxonomy (macro role × Level-of-Autonomy L2–L5) and applies it to existing agentic recommender papers. There is no derivation chain of the kind the circularity analyzer targets: no equations, no fitted parameters renamed as predictions, no uniqueness theorem imported from the authors, and no ansatz smuggled in via self-citation. The LoA spectrum (Figure 1) and three paradigms (Figure 4; §2.3) are author-proposed organizing frames, not results forced by construction from their own inputs. Literature statistics (Figure 2) are descriptive counts of annotated papers, not predictions. Related-survey positioning (Table 2) includes co-author work (e.g., Maragheh & Deldjoo 2025; Deldjoo et al. 2024) in the normal way surveys situate themselves; those citations do not force the taxonomy or the evaluation/open-challenge claims. Component analyses (profile, memory, tools, workflow, optimization) and evaluation discussion (§6) rest on cited third-party systems. Central claim is organizational utility, not a first-principles result that reduces to its premises. Score 0; steps empty.

Axiom & Free-Parameter Ledger

0 free parameters · 4 axioms · 2 invented entities

As a survey, the paper does not fit free parameters to data. Load-bearing content is definitional and taxonomic: what counts as an LLM agent and ARS, the four autonomy dimensions, the L2–L5 focus, and the three macro roles. Those constructs are author-introduced organizing devices applied to external papers; they are not physical entities and have no independent experimental handle outside this framing. Domain assumptions include treating recommendation as an interactive decision process and treating classical RS components as optional tools in agent-as-recommender systems.

axioms (4)
  • domain assumption An LLM agent for recommendation is an entity that observes, reasons, optionally plans, acts (user outputs and/or tools), and updates profile/memory state (Definition 2.1).
    Sets the inclusion boundary for “agentic” vs. single-pass LLM recommenders; invoked throughout §§2–5.
  • ad hoc to paper Autonomy is primarily ordered by proactivity, context awareness, interaction flexibility, and adaptivity, discretized as LoA L2–L5 for concrete systems (§1.1, Figure 1).
    Author-chosen primary axes; alternative orderings (task, modality, domain) are deprioritized.
  • ad hoc to paper Every agentic recommender falls into one of three macro roles: agentic augmentation, replacement, or simulation (§2.3, Figure 4).
    Partition used to structure §§3–5; hybrid systems are forced into the nearest role.
  • domain assumption Recommendation is usefully modeled as a multi-turn interactive decision process with history ht, state st, and action space including dialogue, tool, and environment actions (§2.1).
    Standard sequential decision framing adapted to RS; underpins trajectory-level evaluation claims in §6.
invented entities (2)
  • LoA spectrum L0–L6 for recommender systems no independent evidence
    purpose: Order systems from passive rankers to multi-agent orchestration (and conceptual AGI-level society) and map design ingredients to capabilities and risks.
    Survey-specific ladder; L6 is explicitly conceptual. No falsifiable external prediction beyond classification utility.
  • Two-axis taxonomy (macro role × autonomy level) no independent evidence
    purpose: Cross-classify the agentic RS literature and structure the survey’s analysis of architectures and evaluation.
    Primary organizational invention of the paper; value is interpretive, not experimentally validated against alternative taxonomies.

pith-pipeline@v1.1.0-grok45 · 45696 in / 3134 out tokens · 49741 ms · 2026-07-11T19:12:58.345156+00:00 · methodology

0 comments
read the original abstract

The rapid integration of large language model-based agents into recommender systems has driven a shift from static, ranking-based pipelines toward autonomous and interactive systems that can reason, plan, and act. This survey provides a comprehensive overview of this emerging landscape by introducing a unified taxonomy grounded in the level of autonomy and three core paradigms of agentic recommender systems: agent-assisted recommendation, agent-as-recommender, and agent-as-user-simulator. The autonomy framework organizes existing methods along increasing capabilities in proactivity, context awareness, interaction flexibility, and adaptivity. Building on this framework, the survey analyzes how each paradigm adopts different agentic architectures and how agents enhance key components such as profiles, memory, tool use, workflows, and optimization mechanisms. We further examine evaluation methodologies for agentic recommendation, covering automated metrics, LLM-based judging, and simulation-based assessment, and discuss their limitations in capturing reasoning quality, user experience, and system behavior. Beyond existing evaluation protocols, we further discuss unresolved issues in evaluating agentic recommender systems, including trajectory-level assessment, agent contribution analysis, and calibration of user simulation. Lastly, the survey outlines open challenges in lifelong user modeling, contextual abstraction, multimodal alignment, controllability, trustworthiness, privacy, scalability, and efficiency. Together, these analyses establish a unified foundation for understanding the current progress of agentic recommender systems and highlight promising opportunities for developing more autonomous, reliable, and human-aligned recommendation agents.

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

184 extracted references · 37 linked inside Pith

  1. [1]

    arXiv preprint arXiv:2602.10226 , year=

    Self-evolving recommendation system: End-to-end autonomous model optimization with LLM agents , author=. arXiv preprint arXiv:2602.10226 , year=

    arXiv

  2. [2]

    arXiv:2606.26859 , year=

    AgentX: Towards Agent-Driven Self-Iteration of Industrial Recommender Systems , author=. arXiv:2606.26859 , year=

    Pith/arXiv arXiv

  3. [3]

    Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

    Balancing Fine-tuning and RAG: A Hybrid Strategy for Dynamic LLM Recommendation Updates , author=. Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

  4. [4]

    2025 IEEE Conference on Artificial Intelligence (CAI) , pages=

    Visiorag: A Multimodal Framework for Enhancing Recommendation System Using Vision Transformers and Rag , author=. 2025 IEEE Conference on Artificial Intelligence (CAI) , pages=. 2025 , organization=

    2025

  5. [5]

    IEEE Access , year=

    Leveraging RAG with transformer for context based personalized recommendations , author=. IEEE Access , year=

  6. [6]

    Expert Systems with Applications , pages=

    RALLRec+: Retrieval augmented large language model recommendation with reasoning , author=. Expert Systems with Applications , pages=. 2025 , publisher=

    2025

  7. [7]

    arXiv:2410.19627 , year=

    Knowledge graph enhanced language agents for recommendation , author=. arXiv:2410.19627 , year=

    Pith/arXiv arXiv

  8. [8]

    International Symposium on Practical Aspects of Declarative Languages , pages=

    Automated interactive domain-specific conversational agents that understand human dialogs , author=. International Symposium on Practical Aspects of Declarative Languages , pages=. 2024 , organization=

    2024

  9. [9]

    AAAI , pages=

    RALLM-POI: Retrieval-Augmented LLM for Zero-shot Next POI Recommendation with Geographical Reranking , author=. AAAI , pages=. 2025 , organization=

    2025

  10. [10]

    arXiv preprint arXiv:2510.01609 , year=

    AgentRec: Next-Generation LLM-Powered Multi-Agent Collaborative Recommendation with Adaptive Intelligence , author=. arXiv preprint arXiv:2510.01609 , year=

    arXiv

  11. [11]

    Proceedings of the 34th ACM International Conference on Information and Knowledge Management , pages=

    Temporal-Aware User Behaviour Simulation with Large Language Models for Recommender Systems , author=. Proceedings of the 34th ACM International Conference on Information and Knowledge Management , pages=

  12. [12]

    Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) , pages=

    Evaluating large language models as generative user simulators for conversational recommendation , author=. Proceedings of the 2024 Conference of the North American Chapter of the Association for Computational Linguistics: Human Language Technologies (Volume 1: Long Papers) , pages=

    2024

  13. [13]

    arXiv preprint arXiv:2505.00981 , year=

    Multi-agents based User Values Mining for Recommendation , author=. arXiv preprint arXiv:2505.00981 , year=

    Pith/arXiv arXiv

  14. [14]

    Proceedings of the 18th ACM Conference on Recommender Systems , pages=

    A hybrid multi-agent conversational recommender system with LLM and search engine in e-commerce , author=. Proceedings of the 18th ACM Conference on Recommender Systems , pages=

  15. [15]

    Connecting social media post generation with recommender systems , author=

    C-LINK Agent. Connecting social media post generation with recommender systems , author=. 2024 19th International Workshop on Semantic and Social Media Adaptation & Personalization (SMAP) , pages=. 2024 , organization=

    2024

  16. [16]

    Proceedings of the ACM on Web Conference 2025 , pages=

    A llm-based controllable, scalable, human-involved user simulator framework for conversational recommender systems , author=. Proceedings of the ACM on Web Conference 2025 , pages=

    2025

  17. [17]

    Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems , pages=

    MAP: Multi-user Personalization with Collaborative LLM-powered Agents , author=. Proceedings of the Extended Abstracts of the CHI Conference on Human Factors in Computing Systems , pages=

  18. [18]

    accuracy trade-off: Optimizing rag-enhanced llm recommender systems using multi-head early exit , author=

    The efficiency vs. accuracy trade-off: Optimizing rag-enhanced llm recommender systems using multi-head early exit , author=. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers) , pages=

  19. [19]

    arXiv preprint arXiv:2507.00535 , year=

    Rethinking group recommender systems in the era of generative ai: From one-shot recommendations to agentic group decision support , author=. arXiv preprint arXiv:2507.00535 , year=

    Pith/arXiv arXiv

  20. [20]

    Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining , pages=

    Coral: Collaborative retrieval-augmented large language models improve long-tail recommendation , author=. Proceedings of the 30th ACM SIGKDD conference on knowledge discovery and data mining , pages=

  21. [21]

    Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

    Large Language Model-based Recommendation System Agents , author=. Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

  22. [22]

    arXiv preprint arXiv:2402.01135 , year=

    A multi-agent conversational recommender system , author=. arXiv preprint arXiv:2402.01135 , year=

    Pith/arXiv arXiv

  23. [23]

    Proceedings of the 34th ACM International Conference on Information and Knowledge Management , pages=

    Starec: An efficient agent framework for recommender systems via autonomous deliberate reasoning , author=. Proceedings of the 34th ACM International Conference on Information and Knowledge Management , pages=

  24. [24]

    KDD , year=

    Interactive Recommendation Agent with Active User Commands , author=. KDD , year=

  25. [25]

    Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    iagent: Llm agent as a shield between user and recommender systems , author=. Findings of the Association for Computational Linguistics: ACL 2025 , pages=

    2025

  26. [26]

    Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

    R 4ec: A reasoning, reflection, and refinement framework for recommendation systems , author=. Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

  27. [27]

    2024 IEEE International Conference on Big Data (BigData) , pages=

    Federated recommendation via hybrid retrieval augmented generation , author=. 2024 IEEE International Conference on Big Data (BigData) , pages=. 2024 , organization=

    2024

  28. [28]

    arXiv preprint arXiv:2508.15030 , year=

    Collab-REC: An LLM-based Agentic Framework for Balancing Recommendations in Tourism , author=. arXiv preprint arXiv:2508.15030 , year=

    Pith/arXiv arXiv

  29. [29]

    arXiv preprint arXiv:2508.07672 , year=

    Towards Aligning Personalized Conversational Recommendation Agents with Users' Privacy Preferences , author=. arXiv preprint arXiv:2508.07672 , year=

    Pith/arXiv arXiv

  30. [30]

    Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization , pages=

    Stealthy LLM-driven data poisoning attacks against embedding-based retrieval-augmented recommender systems , author=. Adjunct Proceedings of the 33rd ACM Conference on User Modeling, Adaptation and Personalization , pages=

  31. [31]

    arXiv e-prints , pages=

    Get the Agents Drunk: Memory Perturbations in Autonomous Agent-based Recommender Systems , author=. arXiv e-prints , pages=

  32. [32]

    arXiv:2508.02342 , year=

    Agentic Personalized Fashion Recommendation in the Age of Generative AI: Challenges, Opportunities, and Evaluation , author=. arXiv:2508.02342 , year=

    Pith/arXiv arXiv

  33. [33]

    arXiv preprint arXiv:2506.20817 , year=

    RAG-VisualRec: An Open Resource for Vision-and Text-Enhanced Retrieval-Augmented Generation in Recommendation , author=. arXiv preprint arXiv:2506.20817 , year=

    Pith/arXiv arXiv

  34. [34]

    arXiv preprint arXiv:2510.02668 , year=

    AgenticRAG: Tool-Augmented Foundation Models for Zero-Shot Explainable Recommender Systems , author=. arXiv preprint arXiv:2510.02668 , year=

    arXiv

  35. [35]

    Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

    Lasso: Large Language Model-based User Simulator for Cross-Domain Recommendation , author=. Proceedings of the Nineteenth ACM Conference on Recommender Systems , pages=

  36. [36]

    arXiv preprint arXiv:2601.00930 , year=

    AlignUSER: Human-Aligned LLM Agents via World Models for Recommender System Evaluation , author=. arXiv preprint arXiv:2601.00930 , year=

    arXiv

  37. [37]

    ICML , pages=

    Improving language models by retrieving from trillions of tokens , author=. ICML , pages=. 2022 , organization=

    2022

  38. [38]

    TOIS , volume=

    Recommender ai agent: Integrating large language models for interactive recommendations , author=. TOIS , volume=. 2025 , publisher=

    2025

  39. [39]

    ACL , pages=

    Expectation Confirmation Preference Optimization for Multi-Turn Conversational Recommendation Agent , author=. ACL , pages=

  40. [40]

    WSDM , pages=

    Large language model simulator for cold-start recommendation , author=. WSDM , pages=

  41. [41]

    arXiv:2509.09685 , year=

    Talkplaydata 2: An agentic synthetic data pipeline for multimodal conversational music recommendation , author=. arXiv:2509.09685 , year=

    Pith/arXiv arXiv

  42. [42]

    arXiv preprint arXiv:2507.09188 , year=

    Retrieval-Augmented Recommendation Explanation Generation with Hierarchical Aggregation , author=. arXiv preprint arXiv:2507.09188 , year=

    Pith/arXiv arXiv

  43. [43]

    ACM Transactions on Information Systems , volume=

    User behavior simulation with large language model-based agents , author=. ACM Transactions on Information Systems , volume=. 2025 , publisher=

    2025

  44. [44]

    arXiv:2501.02727 , year=

    Tree-based RAG-Agent Recommendation System: A Case Study in Medical Test Data , author=. arXiv:2501.02727 , year=

    Pith/arXiv arXiv

  45. [45]

    Proceedings of the AAAI Conference on Artificial Intelligence , volume=

    Llm-powered user simulator for recommender system , author=. Proceedings of the AAAI Conference on Artificial Intelligence , volume=

  46. [46]

    Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    LLM as User Simulator: Towards Training News Recommender without Real User Interactions , author=. Proceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  47. [47]

    arXiv:2506.17765 , year=

    CARTS: Collaborative Agents for Recommendation Textual Summarization , author=. arXiv:2506.17765 , year=

    Pith/arXiv arXiv

  48. [48]

    arXiv:2306.02552 , year=

    Recagent: A novel simulation paradigm for recommender systems , author=. arXiv:2306.02552 , year=

    Pith/arXiv arXiv

  49. [49]

    AAAI , year=

    Diagnostic-Guided Dynamic Profile Optimization for LLM-based User Simulators in Sequential Recommendation , author=. AAAI , year=

  50. [50]

    Companion Proceedings of the ACM on Web Conference 2025 , pages=

    RecUserSim: A Realistic and Diverse User Simulator for Evaluating Conversational Recommender Systems , author=. Companion Proceedings of the ACM on Web Conference 2025 , pages=

    2025

  51. [51]

    Proceedings of the 48th International ACM SIGIR conference on Research and Development in Information Retrieval , pages=

    Agentic feedback loop modeling improves recommendation and user simulation , author=. Proceedings of the 48th International ACM SIGIR conference on Research and Development in Information Retrieval , pages=

  52. [52]

    arXiv preprint arXiv:2509.10397 , year=

    Recoworld: Building simulated environments for agentic recommender systems , author=. arXiv preprint arXiv:2509.10397 , year=

    arXiv

  53. [53]

    arXiv preprint arXiv:2504.08742 , year=

    Simulating Filter Bubble on Short-video Recommender System with Large Language Model Agents , author=. arXiv preprint arXiv:2504.08742 , year=

    Pith/arXiv arXiv

  54. [54]

    arXiv:2505.21154 , year=

    GGBond: Growing Graph-Based AI-Agent Society for Socially-Aware Recommender Simulation , author=. arXiv:2505.21154 , year=

    Pith/arXiv arXiv

  55. [55]

    AI Agent for Information Retrieval Workshop , pages=

    Agent0: Leveraging LLM Agents to Discover Multi-value Features from Text for Enhanced Recommendations , author=. AI Agent for Information Retrieval Workshop , pages=. 2024 , organization=

    2024

  56. [56]

    Proceedings of the ACM Web Conference 2024 , pages=

    Agentcf: Collaborative learning with autonomous language agents for recommender systems , author=. Proceedings of the ACM Web Conference 2024 , pages=

    2024

  57. [57]

    arXiv:2510.13371 , year=

    MADREC: A Multi-Aspect Driven LLM Agent for Explainable and Adaptive Recommendation , author=. arXiv:2510.13371 , year=

    arXiv

  58. [58]

    arXiv preprint arXiv:2507.02626 , year=

    VRAgent-R1: Boosting Video Recommendation with MLLM-based Agents via Reinforcement Learning , author=. arXiv preprint arXiv:2507.02626 , year=

    Pith/arXiv arXiv

  59. [59]

    arXiv preprint arXiv:2508.13423 , year=

    AdaptJobRec: Enhancing Conversational Career Recommendation through an LLM-Powered Agentic System , author=. arXiv preprint arXiv:2508.13423 , year=

    arXiv

  60. [60]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    Retrieval-augmented conversational recommendation with prompt-based semi-structured natural language state tracking , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  61. [61]

    Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

    Macrec: A multi-agent collaboration framework for recommendation , author=. Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval , pages=

  62. [62]

    arXiv:2510.06078 , year=

    Constraint-Aware Route Recommendation from Natural Language via Hierarchical LLM Agents , author=. arXiv:2510.06078 , year=

    arXiv

  63. [63]

    Proceedings of the 7th ACM Conference on Conversational User Interfaces , pages=

    SmartEats: Investigating the Effects of Customizable Conversational Agent in Dietary Recommendations , author=. Proceedings of the 7th ACM Conference on Conversational User Interfaces , pages=

  64. [64]

    arXiv:2509.01551 , year=

    Cloud-Device Collaborative Agents for Sequential Recommendation , author=. arXiv:2509.01551 , year=

    Pith/arXiv arXiv

  65. [65]

    arXiv:2308.10835 , year=

    Enhancing recommender systems with large language model reasoning graphs , author=. arXiv:2308.10835 , year=

    Pith/arXiv arXiv

  66. [66]

    2025 , month = dec, type =

    Conversational Recommender Systems Using Generative Models (Gen-CRS): A Literature Review , author =. 2025 , month = dec, type =

    2025

  67. [67]

    2018 IEEE international conference on data mining (ICDM) , pages=

    Self-attentive sequential recommendation , author=. 2018 IEEE international conference on data mining (ICDM) , pages=. 2018 , organization=

    2018

  68. [68]

    Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence , pages=

    BPR: Bayesian personalized ranking from implicit feedback , author=. Proceedings of the Twenty-Fifth Conference on Uncertainty in Artificial Intelligence , pages=

  69. [69]

    arXiv:2507.02097 , year=

    The Future is Agentic: Definitions, Perspectives, and Open Challenges of Multi-Agent Recommender Systems , author=. arXiv:2507.02097 , year=

    Pith/arXiv arXiv

  70. [70]

    2025 , month = apr, url =

    A Practical Guide to Building Agents , institution =. 2025 , month = apr, url =

    2025

  71. [71]

    arXiv preprint arXiv:2506.21931 , year=

    ARAG: Agentic Retrieval Augmented Generation for Personalized Recommendation , author=. arXiv preprint arXiv:2506.21931 , year=

    Pith/arXiv arXiv

  72. [72]

    arXiv preprint arXiv:2504.20094 , year=

    MATCHA: Can Multi-Agent Collaboration Build a Trustworthy Conversational Recommender? , author=. arXiv preprint arXiv:2504.20094 , year=

    arXiv

  73. [73]

    World Wide Web , volume=

    A survey on large language models for recommendation , author=. World Wide Web , volume=. 2024 , publisher=

    2024

  74. [74]

    arXiv preprint arXiv:2503.05659 , year=

    A survey of large language model empowered agents for recommendation and search: Towards next-generation information retrieval , author=. arXiv preprint arXiv:2503.05659 , year=

    Pith/arXiv arXiv

  75. [75]

    arXiv preprint arXiv:2503.16734 , year=

    Towards agentic recommender systems in the era of multimodal large language models , author=. arXiv preprint arXiv:2503.16734 , year=

    Pith/arXiv arXiv

  76. [76]

    arXiv preprint arXiv:2502.10050 , year=

    A survey on llm-powered agents for recommender systems , author=. arXiv preprint arXiv:2502.10050 , year=

    Pith/arXiv arXiv

  77. [77]

    arXiv preprint arXiv:2504.16420 , year=

    A Survey of Foundation Model-Powered Recommender Systems: From Feature-Based, Generative to Agentic Paradigms , author=. arXiv preprint arXiv:2504.16420 , year=

    Pith/arXiv arXiv

  78. [78]

    Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    Large language models for generative recommendation: A survey and visionary discussions , author=. Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024) , pages=

    2024

  79. [79]

    IEEE Transactions on Knowledge and Data Engineering , volume=

    Recommender systems in the era of large language models (llms) , author=. IEEE Transactions on Knowledge and Data Engineering , volume=. 2024 , publisher=

    2024

  80. [80]

    Foundations and Trends

    Recommender systems meet large language model agents: A survey , author=. Foundations and Trends. 2025 , publisher=

    2025

Showing first 80 references.