Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

The paper proposes ACERec, which distills 32-token OPQ semantic IDs into four latent tokens via an Attentive Token Merger, and reports a 14.40% average NDCG@10 gain over the strongest generative-recommendation baselines.

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 →

T0 review

2026-08-04 06:00 UTC pith:ENX5EPPZ

load-bearing objection Solid architectural contribution to OPQ-based generative recommendation, but the headline accuracy numbers need variance and reporting fixes before they can be trusted. the 4 major comments →

arxiv 2602.13573 v2 pith:ENX5EPPZ submitted 2026-02-14 cs.IR

Unleash the Potential of Long Semantic IDs for Generative Recommendation

classification cs.IR
keywords semantic IDsgenerative recommendationsequential recommendationoptimized product quantizationattentive token mergerintent tokendual-granularity optimizationcold-start recommendation
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.

Generative recommenders represent items as sequences of discrete semantic tokens, but efficient models are forced to use short IDs that lose fine-grained item semantics, while long IDs that capture those semantics make sequence modeling computationally prohibitive. ACERec claims to break this trade-off by decoupling tokenization granularity from modeling cost: it builds 32-token semantic IDs with orthogonal product quantization, then uses an Attentive Token Merger to distill each item into four compact latent tokens before sequential modeling. A learnable Intent Token, supervised by a dual-granularity objective that combines token-level multi-token prediction with item-level semantic alignment, anchors next-item generation. Across six Amazon benchmarks, the paper reports that ACERec consistently outperforms state-of-the-art baselines, with an average NDCG@10 improvement of 14.40% and especially strong cold-start gains. If the claim holds, long semantic IDs become practical for real recommender systems, preserving expressiveness at a fraction of the compute.

Core claim

The central claim is that OPQ-based recommenders lose semantic fidelity not because long IDs are inherently wasteful but because of how they are consumed: pooling blurs subspace attributes, and raw sequences are too costly. ACERec interposes an Attentive Token Merger that compresses each item's 32 OPQ tokens into four latent tokens via cross-attention with item-specific queries, preserving salient subspace signals. A learnable Intent Token, updated under step-wise causal attention, becomes the prediction anchor, and a dual-granularity objective — multi-token prediction plus popularity-debiased item-level semantic alignment — supervises it. The paper reports average NDCG@10 gains of 14.40% ov

What carries the argument

OPQ (Optimized Product Quantization) is the tokenizer: it decomposes each item embedding into m orthogonal subspaces, yielding long discrete semantic IDs. The core is the Attentive Token Merger (ATM): content-adaptive latent queries (k=4) cross-attend to the m=32 token embeddings and distill them into compact latents, preserving subspace identity. A learnable Intent Token per step, updated under step-wise causal attention, becomes the dynamic prediction anchor. A dual-granularity objective combines multi-token prediction (parallel, bit-accurate) with a popularity-debiased item-level semantic alignment (ISA), yielding both token exactness and holistic intent grounding. Holistic candidate scor

Load-bearing premise

Everything hinges on the premise that splitting items into many orthogonal quantized subspaces yields semantically meaningful dimensions that the attention-based merger can preserve; if those subspaces are arbitrary or the merger discards the crucial ones, the long-ID advantage disappears.

What would settle it

Train a linear probe on the fused latent tokens (k=4) to predict known item attributes (category, brand, color) and compare it against the same probe on mean-pooled OPQ embeddings of the same items; if the probe on ATM latents does not outperform the probe on pooled embeddings, the claimed semantic preservation collapses, because any ranking gain would then have to come from something other than fine-grained attribute retention.

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

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Long semantic IDs (m=32) become usable in production: the recommender only sees k+1 tokens per item, so the quadratic attention cost is cut by a factor of roughly (k/m)^2 relative to raw long-ID modeling.
  • Cold-start and long-tail items gain the most: the paper reports that ACERec nearly doubles the NDCG@10 of the best pooling-based baseline on items with 0–5 training interactions, consistent with knowledge transfer through shared semantic tokens.
  • The dual-granularity objective is a general recipe: combining bit-accurate token prediction with item-level semantic alignment gives a better prediction anchor than static pooling of a Transformer's last hidden state.
  • Exact retrieval over the full catalog becomes cheap: candidate scoring is a memory gather of precomputed codebook log-probabilities, with cost independent of corpus size, removing the need for approximate nearest-neighbor decoding.
  • The decoupling principle—tokenization resolution should be chosen for semantic expressiveness, not for the sequence model's budget—suggests that even longer IDs (e.g., 64) can be used when domains demand finer distinctions.

Where Pith is reading between the lines

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

  • The optimal compression ratio (default r=8) likely depends on catalog diversity and how well OPQ subspaces align with real attributes; the paper's own data show Instruments prefers m=64, so a data-driven schedule for m and k is a natural next step that the paper does not take.
  • Because ATM's queries are conditioned only on the item, the same compression is applied regardless of user context; making the queries user- or session-dependent could personalize which semantic subspaces are emphasized, a variant not explored here.
  • The assumption that OPQ's orthogonal subspaces are semantically disentangled is probably only approximate. If the merger is truly content-adaptive, the same architecture could be trained end-to-end with a learnable quantizer, potentially closing the gap further—though the paper fixes OPQ to keep the comparison fair.
  • The idea of distilling long discrete codes into compact latents before a sequence model may transfer beyond recommendation to any generative retrieval task over quantized codes, such as document or image retrieval, where the same granularity-efficiency wall exists.

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

4 major / 4 minor

Summary. The paper proposes ACERec, a generative sequential recommender that uses OPQ-based long semantic IDs (m=32/64) but compresses them into k=4 latent tokens via an Attentive Token Merger (ATM), avoiding both the serial-decoding cost of RQ and the semantic blurring of rigid mean pooling. It adds an Intent Token as a dynamic prediction anchor and trains with a dual-granularity objective combining multi-token prediction (MTP) with an intent–semantic alignment (ISA) contrastive loss. Inference is exact and fully parallel, performed by holistic candidate scoring with vectorized gathering of subspace log-probabilities. Experiments on six Amazon datasets report consistent improvements over item-ID and semantic-ID baselines, with an average 14.40% NDCG@10 gain in the full-text abstract (the metadata abstract instead reports 12.92% NDCG@10 and 7.49% Recall@10 on nine benchmarks). Additional analyses cover cold-start behavior, compression ratio, inference efficiency, and convergence.

Significance. If the empirical claims hold, ACERec would be a practically valuable upgrade over RPG and TIGER: it uses long expressive OPQ IDs while keeping the sequential input short, it avoids beam search and graph decoding, and its exact candidate scoring is independent of catalog size. The architecture is clean and the ablations isolate the contributions of ATM, the Intent Token, and ISA. The paper also ships a clearly specified inference algorithm and makes a serious effort to use a unified semantic encoder for semantic-ID methods. The main weakness is that the central empirical assertion—consistent state-of-the-art performance—rests on single runs without variance or significance tests, on an inconsistent dataset count between the two abstract versions, and on borrowed baseline numbers for three datasets. The semantic-preservation premise of ATM is also validated only through downstream ranking, not through an independent fidelity or disentanglement measure.

major comments (4)
  1. [Abstract / §3.1] The full-text abstract states 'six real-world benchmarks' and 'an average improvement of 14.40% in NDCG@10', while the metadata abstract states 'nine real-world benchmarks' with '12.92% in NDCG@10 and 7.49% in Recall@10'. The body reports experiments on only six datasets (Sports, Beauty, Toys, Instruments, Office, Baby). This is not a cosmetic issue: the headline performance claim is numerically ambiguous. Please reconcile the two abstract versions, state exactly which datasets and which baseline the average improvement is computed over, and provide the per-dataset gains.
  2. [Tables 1–2 and Appendix A.2] The claim that ACERec 'consistently outperforms state-of-the-art baselines' is not statistically secured. Tables 1 and 2 report a single run without error bars, confidence intervals, or multiple seeds. Several gains are small, notably Recall@10 on Office (+3.44%) and Baby (+0.71%), which could plausibly fall within run-to-run noise. In addition, Appendix A.2 states that for Sports, Beauty, and Toys the results of HGN, SASRec, S3-Rec, and TIGER are adopted from Rajput et al., while the same appendix says a unified sentence-t5-base encoder is used for all semantic-ID methods. A TIGER result from the original paper may use a different semantic encoder and data split, so the comparison may not be controlled. Please reproduce these baselines with the same protocol or justify the compatibility.
  3. [Table 3 and §3.3.1] The ablation claim that ISA is effective independently of ATM is only supported on Instruments. On Office, adding ISA without ATM (Row 3 vs. Row 4) degrades all four metrics: Recall@5 drops 0.0640→0.0626, NDCG@5 0.0433→0.0421, Recall@10 0.0993→0.0991, and NDCG@10 0.0546→0.0537. The text says 'adding the LISA objective to the Intent Token model brings a clear gain' and cites the Instruments example, but the Office result contradicts this. Please describe the result more carefully and discuss the inconsistency.
  4. [§2.2.1 and §3.4.2] The framework presupposes that OPQ's orthogonal subspaces 'largely disentangle semantic attributes across digits' and that longer IDs are semantically more expressive. This is supported only through downstream benchmark improvements, not by an independent measurement of semantic fidelity. Because ATM's 'faithfulness' is evaluated by the same ranking objective used to train it, the semantic-preservation interpretation is partly circular. Please add a direct test—for example, reconstruction fidelity of the merged latents, attribute probing, or a comparison of ATM's output against the original long IDs on a non-ranking semantic task—to substantiate the claim that ATM preserves fine-grained information beyond what short IDs provide.
minor comments (4)
  1. [Figures 3, 4, 10, 12] Several figures lack axis labels or legends. In Figure 3, the x-axis appears to be the compression ratio r but is not labeled; Figures 4, 10, and 12 also omit y-axis units and legend details. This makes the quantitative claims hard to verify from the figures alone.
  2. [Equation (1)] The aggregation function f_s(e_1, ..., e_m) is not defined. It should be stated explicitly whether this is mean pooling, a permutation-invariant aggregation, or a learned attention pooling over the token embeddings, and how the positional embeddings P are incorporated before this aggregation.
  3. [§3.4.2] The claimed average improvements of 44.48% in Recall@10 and 56.91% in NDCG@10 over the short-digit OPQ baseline appear only in the text and are not tied to a table. Please report per-dataset numbers (or an appendix table) so the reader can assess the consistency of this gain.
  4. [Table 1] The 'Improv.' row is computed against the strongest baseline for each metric, but §3.2 highlights 'large improvements over RPG' on Instruments. For Instruments, the strongest baseline for NDCG@10 is ELCRec, not RPG. Please specify the reference baseline in each place to avoid ambiguity.

Circularity Check

0 steps flagged

No significant circularity: benchmark gains are measured, and ATM/Intent/ISA are trained components rather than fitted constants relabeled as predictions.

full rationale

ACERec's load-bearing claims are empirical: the reported NDCG@10/Recall@10 gains are measured benchmark outcomes, not numbers recovered from fitted constants. The method's components—ATM compression (Eq. 2), the Intent Token, the MTP loss (Eq. 4), and the ISA loss (Eq. 6)—are new trained modules operating on fixed OPQ codebooks; inference scoring (Eqs. 8–9) is the standard generative-retrieval gather over predicted codebook log-probabilities, so the final ranking is the model's output distribution rather than a re-display of its input. The cited machinery (OPQ/FAISS, MTP, logit adjustment) is external, parameter-free with stated assumptions, and not a self-citation chain. The only caveats—baseline numbers adopted from Rajput et al. in Appendix A.2, absence of error bars, and the fact that ATM's 'semantic fidelity' is inferred from downstream/attention evidence rather than an independent fidelity metric—are external-validity or reporting concerns, not definitional or fitted-input circularity. No uniqueness theorem or self-citation is used to force the design.

Axiom & Free-Parameter Ledger

7 free parameters · 5 axioms · 2 invented entities

The central method is not derived from first principles; it is assembled from standard components and tuned hyperparameters. The only genuinely new entities are architectural (ATM, Intent Token), and their value is established solely through benchmark performance within the same framework.

free parameters (7)
  • ISA weight lambda = tuned per dataset in {0.03, 0.05, 0.07, 0.1} after initial 0.01
    Controls the contribution of the item-level alignment loss; selected by validation NDCG@10 in a 13-run tuning protocol (Appendix A.2).
  • Popularity debiasing strength beta = 0.02
    Set by hand in Eq. 5 to correct in-batch sampling bias; affects ISA gradients and final scores.
  • MTP temperature gamma = 0.03
    Temperature in codebook softmax (Eqs. 3 and 8); chosen by hand.
  • ISA temperature tau = 0.07
    Temperature in the debiased contrastive score (Eq. 5); chosen by hand.
  • Compression ratio r and latent count k = r=8, so k=4 for m=32; k=8 for m=64 on Instruments
    Selected from Figure 3 as the 'optimal semantic filter'; not derived from first principles.
  • Semantic ID length m = 32 for most datasets; 64 for Instruments
    Selected per dataset from validation curves (Figure 7); a tuned input representation choice.
  • Codebook size M = 256
    Vocabulary size per OPQ subspace; fixed but hand-chosen.
axioms (5)
  • domain assumption OPQ orthogonal subspace decomposition semantically disentangles item attributes and permits conditional independence across token digits.
    Invoked in Section 2.2.1 to justify parallel prediction and the value of long IDs; no independent semantic-disentanglement evidence is provided.
  • domain assumption Sentence-T5 encodes item text into an embedding space that serves as a valid semantic basis for recommendation.
    Appendix A.2 uses sentence-t5-base as the universal semantic encoder for all semantic-ID methods; if these embeddings are poor for the domain, all compared methods suffer and the premise of semantic IDs weakens.
  • domain assumption Leave-last-out split with chronological ordering and full-candidate ranking is a faithful evaluation of sequential recommendation quality.
    Section 3.1.1; standard in the literature but a simplification of real deployment (no exposure or serving bias).
  • domain assumption The logit-adjustment term -beta log P(j) corrects in-batch popularity bias.
    Eq. 5; the correction depends on empirical sampling probabilities and a tuned beta; no analysis shows it removes bias in this setting.
  • standard math Standard Transformer/attention machinery and backpropagation behave as expected.
    Background for Eqs. 1-9; not proved in the paper but accepted in the field.
invented entities (2)
  • Attentive Token Merger (ATM) no independent evidence
    purpose: Compresses m=32 token embeddings into k=4 latent tokens via content-adaptive cross-attention.
    An internal model component; its 'semantic preservation' is verified only through downstream recommendation metrics, not through any external fidelity test.
  • Intent Token no independent evidence
    purpose: A learnable per-step token that aggregates historical latents and serves as the prediction anchor for next-item generation.
    Model-internal state; no falsifiable handle outside the trained system.

reviewed 2026-08-04 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Unleash the Potential of Long Semantic IDs for Generative Recommendation." pith.science (2026). https://pith.science/paper/ENX5EPPZ

@misc{pith2026260213573,
  author       = {Pith},
  title        = {Pith review of: Unleash the Potential of Long Semantic IDs for Generative Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ENX5EPPZ}},
  note         = {Machine review of arXiv:2602.13573}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Semantic ID-based generative recommenders face a granularity-efficiency dilemma between efficient recommendation with short IDs and expressive item modeling with long IDs. To break this dilemma, we propose ACERec, a framework that preserves the semantic richness of long IDs while keeping the recommendation process efficient. Concretely, ACERec employs an Attentive Token Merger to compress long semantic IDs into compact yet faithful latent tokens. To better capture user intent from the compressed semantics, we further introduce a dedicated Intent Token, optimized by a dual-granularity objective that combines token-level generation with item-level intent-semantic alignment. Extensive experiments on nine real-world benchmarks show that ACERec consistently outperforms state-of-the-art methods, yielding average relative improvements of 12.92% in NDCG@10 and 7.49% in Recall@10 over the strongest baselines.

Figures

Figures reproduced from arXiv: 2602.13573 by Dongmin Huang, Guoxin Ma, Ming Xia, Zhiqin Zhou.

Figure 1
Figure 1. Figure 1: Comparison of representation paradigms. Left: RQ [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview architecture of ACERec. The left panel illustrates the sequence encoding process, where historical items [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Figure 4: Performance comparison between ACERec and [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 3
Figure 3. Figure 3: , performance exhibits a distinct inverted V-shaped trend, peaking at 𝑟 = 8. Notably, lower compression ratios (𝑟=2, 4) underperform com￾pared to 𝑟=8, suggesting that excessive semantic detail may overwhelm the sequence model and dilute important information. Conversely, overly aggressive compression (𝑟=16) creates an infor￾mation bottleneck, causing significant performance drops. Thus, 𝑟=8 serves as an op… view at source ↗
Figure 5
Figure 5. Figure 5: Cold-start analysis on Instruments and Baby [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Inference Efficiency and Recommendation Perfor [PITH_FULL_IMAGE:figures/full_fig_p008_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: NDCG@10 performance under different digit length [PITH_FULL_IMAGE:figures/full_fig_p012_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Performance comparison of different loss coeffi [PITH_FULL_IMAGE:figures/full_fig_p012_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Additional cold-start performance (NDCG@10) on Beauty, Toys, and Office datasets. [PITH_FULL_IMAGE:figures/full_fig_p013_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: NDCG@10 Performance Comparison with varying [PITH_FULL_IMAGE:figures/full_fig_p013_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Visualization of ATM compression [PITH_FULL_IMAGE:figures/full_fig_p014_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Training Dynamics and Convergence Rate on Toys [PITH_FULL_IMAGE:figures/full_fig_p014_12.png] view at source ↗

discussion (0)

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

Forward citations

Cited by 1 Pith paper

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

  1. SIDInspector: A Mapping-First Diagnostic Resource for Semantic-ID Tokenizers

    cs.IR 2026-06 accept novelty 6.0

    SIDInspector provides a standardized adapter contract and mapping-level probes for Semantic-ID tokenizers, with empirical contrasts showing high aliasing in GRID-style exports and superior prefix alignment from determ...

Reference graph

Works this paper leans on

43 extracted references · 6 linked inside Pith · cited by 1 Pith paper

  1. [1]

    Prabhat Agarwal, Anirudhan Badrinath, Laksh Bhasin, Jaewon Yang, Edoardo Botta, Jiajing Xu, and Charles Rosenberg. 2025. Pinrec: Outcome-conditioned, multi-token generative retrieval for industry-scale recommendation systems. arXiv preprint arXiv:2504.10507(2025)

  2. [2]

    Yoshua Bengio and Jean-Sébastien Senécal. 2008. Adaptive importance sampling to accelerate training of a neural probabilistic language model.IEEE Transactions on Neural Networks19, 4 (2008), 713–722

  3. [3]

    Yukuo Cen, Jianwei Zhang, Xu Zou, Chang Zhou, Hongxia Yang, and Jie Tang

  4. [4]

    Yongjun Chen, Zhiwei Liu, Jia Li, Julian McAuley, and Caiming Xiong. 2022. Intent contrastive learning for sequential recommendation. InProceedings of the ACM web conference 2022. 2172–2182

  5. [5]

    Matthijs Douze, Alexandr Guzhva, Chengqi Deng, Jeff Johnson, Gergely Szilvasy, Pierre-Emmanuel Mazaré, Maria Lomeli, Lucas Hosseini, and Hervé Jégou. 2025. The faiss library.IEEE Transactions on Big Data(2025)

  6. [6]

    Tiezheng Ge, Kaiming He, Qifa Ke, and Jian Sun. 2013. Optimized product quantization.IEEE transactions on pattern analysis and machine intelligence36, 4 (2013), 744–755

  7. [7]

    Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022. Recommendation as language processing (rlp): A unified pretrain, personalized prompt & predict paradigm (p5). InProceedings of the 16th ACM conference on recommender systems. 299–315

  8. [8]

    Fabian Gloeckle, Badr Youbi Idrissi, Baptiste Rozière, David Lopez-Paz, and Gabriel Synnaeve. 2024. Better & faster large language models via multi-token prediction.arXiv preprint arXiv:2404.19737(2024)

  9. [9]

    Balázs Hidasi, Alexandros Karatzoglou, Linas Baltrunas, and Domonkos Tikk

  10. [10]

    Chris Hokamp and Qun Liu. 2017. Lexically constrained decoding for sequence generation using grid beam search.arXiv preprint arXiv:1704.07138(2017)

  11. [11]

    Yupeng Hou, Zhankui He, Julian McAuley, and Wayne Xin Zhao. 2023. Learning vector-quantized item representation for transferable sequential recommenders. InProceedings of the ACM Web Conference 2023. 1162–1171

  12. [12]

    Yupeng Hou, Jiacheng Li, Ashley Shin, Jinsung Jeon, Abhishek Santhanam, Wei Shao, Kaveh Hassani, Ning Yao, and Julian McAuley. 2025. Generating long semantic ids in parallel for recommendation. InProceedings of the 31st ACM SIGKDD Conference on Knowledge Discovery and Data Mining V. 2. 956–966

  13. [13]

    Yupeng Hou, Jianmo Ni, Zhankui He, Noveen Sachdeva, Wang-Cheng Kang, Ed H Chi, Julian McAuley, and Derek Zhiyuan Cheng. 2025. ActionPiece: Contextually Tokenizing Action Sequences for Generative Recommendation.arXiv preprint arXiv:2502.13581(2025)

  14. [14]

    Yupeng Hou, An Zhang, Leheng Sheng, Jiancan Wu, Xiang Wang, Tat-Seng Chua, and Julian McAuley. 2025. Towards large generative recommendation: A tokenization perspective. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 6821–6824

  15. [15]

    Herve Jegou, Matthijs Douze, and Cordelia Schmid. 2010. Product quantization for nearest neighbor search.IEEE transactions on pattern analysis and machine intelligence33, 1 (2010), 117–128

  16. [16]

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

  17. [17]

    Chao Li, Zhiyuan Liu, Mengmeng Wu, Yuchi Xu, Huan Zhao, Pipei Huang, Guoliang Kang, Qiwei Chen, Wei Li, and Dik Lun Lee. 2019. Multi-interest network with dynamic routing for recommendation at Tmall. InProceedings of the 28th ACM international conference on information and knowledge management. 2615–2623

  18. [18]

    Jing Li, Pengjie Ren, Zhumin Chen, Zhaochun Ren, Tao Lian, and Jun Ma. 2017. Neural attentive session-based recommendation. InProceedings of the 2017 ACM on Conference on Information and Knowledge Management. 1419–1428

  19. [19]

    Lei Li, Yongfeng Zhang, Dugang Liu, and Li Chen. 2024. Large language models for generative recommendation: A survey and visionary discussions. In Proceedings of the 2024 Joint International Conference on Computational Linguistics, Language Resources and Evaluation (LREC-COLING 2024). 10146–10159

  20. [20]

    Enze Liu, Bowen Zheng, Cheng Ling, Lantao Hu, Han Li, and Wayne Xin Zhao

  21. [21]

    Yue Liu, Shihao Zhu, Jun Xia, Yingwei Ma, Jian Ma, Xinwang Liu, Shengju Yu, Kejun Zhang, and Wenliang Zhong. 2024. End-to-end learnable clustering for intent learning in recommendation.Advances in Neural Information Processing Systems37 (2024), 5913–5949

  22. [22]

    Chen Ma, Peng Kang, and Xue Liu. 2019. Hierarchical gating networks for sequential recommendation. InProceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining. 825–833

  23. [23]

    Julian McAuley, Christopher Targett, Qinfeng Shi, and Anton Van Den Hengel

  24. [24]

    Jianmo Ni, Gustavo Hernandez Abrego, Noah Constant, Ji Ma, Keith Hall, Daniel Cer, and Yinfei Yang. 2022. Sentence-t5: Scalable sentence encoders from pre- trained text-to-text models. InFindings of the association for computational linguistics: ACL 2022. 1864–1874

  25. [25]

    Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al

  26. [26]

    Steffen Rendle, Christoph Freudenthaler, and Lars Schmidt-Thieme. 2010. Factorizing personalized markov chains for next-basket recommendation. In Proceedings of the 19th international conference on World wide web. 811–820

  27. [27]

    InProceedings of the 38th international ACM SIGIR conference on research and development in information retrieval

    Image-based recommendations on styles and substitutes. InProceedings of the 38th international ACM SIGIR conference on research and development in information retrieval. 43–52

  28. [28]

    Jiaxi Tang and Ke Wang. 2018. Personalized top-n sequential recommendation via convolutional sequence embedding. InProceedings of the eleventh ACM international conference on web search and data mining. 565–573

  29. [29]

    Ashish Vaswani, Noam Shazeer, Niki Parmar, Jakob Uszkoreit, Llion Jones, Aidan N Gomez, Łukasz Kaiser, and Illia Polosukhin. 2017. Attention is all you need.Advances in neural information processing systems30 (2017)

  30. [30]

    Hao Wang, Wei Guo, Luankang Zhang, Jin Yao Chin, Yufei Ye, Huifeng Guo, Yong Liu, Defu Lian, Ruiming Tang, and Enhong Chen. 2025. Generative large recommendation models: Emerging trends in llms for recommendation. In Companion Proceedings of the ACM on Web Conference 2025. 49–52

  31. [31]

    Wenjie Wang, Xinyu Lin, Fuli Feng, Xiangnan He, and Tat-Seng Chua. 2023. Generative recommendation: Towards next-generation recommender paradigm. arXiv preprint arXiv:2304.03516(2023)

  32. [32]

    Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang. 2019. BERT4Rec: Sequential recommendation with bidirectional encoder representations from transformer. InProceedings of the 28th ACM international conference on information and knowledge management. 1441–1450

  33. [33]

    Xinyang Yi, Ji Yang, Lichan Hong, Derek Zhiyuan Cheng, Lukasz Heldt, Aditee Kumthekar, Zhe Zhao, Li Wei, and Ed Chi. 2019. Sampling-bias-corrected neural modeling for large corpus item recommendations. InProceedings of the 13th ACM conference on recommender systems. 269–277

  34. [34]

    Zhenrui Yue, Yueqi Wang, Zhankui He, Huimin Zeng, Julian McAuley, and Dong Wang. 2024. Linear recurrent units for sequential recommendation. In Proceedings of the 17th ACM international conference on web search and data mining. 930–938

  35. [35]

    Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. 2021. Soundstream: An end-to-end neural audio codec.IEEE/ACM Transactions on Audio, Speech, and Language Processing30 (2021), 495–507

  36. [36]

    Tingting Zhang, Pengpeng Zhao, Yanchi Liu, Victor S Sheng, Jiajie Xu, Deqing Wang, Guanfeng Liu, Xiaofang Zhou, et al . 2019. Feature-level deeper self- attention network for sequential recommendation.. InIJCAI. 4320–4326

  37. [37]

    Thomas Wolf, Lysandre Debut, Victor Sanh, Julien Chaumond, Clement Delangue, Anthony Moi, Pierric Cistac, Tim Rault, Remi Louf, Morgan Funtowicz, et al . 2020. Transformers: State-of-the-art natural language processing. In Proceedings of the 2020 conference on empirical methods in natural language processing: system demonstrations. 38–45

  38. [38]

    sweet spot

    Kun Zhou, Hui Wang, Wayne Xin Zhao, Yutao Zhu, Sirui Wang, Fuzheng Zhang, Zhongyuan Wang, and Ji-Rong Wen. 2020. S3-rec: Self-supervised learning for sequential recommendation with mutual information maximization. In Proceedings of the 29th ACM international conference on information & knowledge management. 1893–1902. Unleash the Potential of Long Semanti...

  39. [42]

    Wayne Xin Zhao, Shanlei Mu, Yupeng Hou, Zihan Lin, Yushuo Chen, Xingyu Pan, Kaiyuan Li, Yujie Lu, Hui Wang, Changxin Tian, et al. 2021. Recbole: Towards a unified, comprehensive and efficient framework for recommendation algorithms. Inproceedings of the 30th acm international conference on information & knowledge management. 4653–4664

  40. [2015]

    Session-based recommendations with recurrent neural networks.arXiv preprint arXiv:1511.06939(2015)

  41. [2020]

    InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining

    Controllable multi-interest framework for recommendation. InProceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining. 2942–2951

  42. [2023]

    Recommender systems with generative retrieval.Advances in Neural Information Processing Systems36 (2023), 10299–10315

  43. [2025]

    InProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval

    Generative recommender with end-to-end learnable item tokenization. InProceedings of the 48th International ACM SIGIR Conference on Research and Development in Information Retrieval. 729–739

This paper was first reviewed by deepseek-v4-flash on August 4, 2026.