LLM-RecG: A Semantic Bias-Aware Framework for Zero-Shot Sequential Recommendation

Hari Sundaram; Junting Wang; Yunzhe Li; Zhining Liu

REVIEW 3 major objections 5 minor 2 cited by

LLM-RecG: A Semantic Bias-Aware Framework for Zero-Shot Sequential Recommendation

T0 review · 3 major / 5 minor · reviewed 2026-08-09 · deepseek-v4-flash

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

Pith's one-line read Domain semantic bias in LLM embeddings is the main barrier to zero-shot cross-domain sequential recommendation, and a dual-level generalization loss overcomes it.

desk verdict The paper's central inter-domain compactness loss is identically zero in every two-domain experiment as written, and the zero-shot protocol leaks target-domain information; the empirical work is substantial but the formal claims do not hold. read the letter →

arxiv 2501.19232 v2 pith:7EH4EUVC submitted 2025-01-31 cs.IR cs.AI

Yunzhe Li , Junting Wang , Hari Sundaram , Zhining Liu This is my paper · ORCID

classification cs.IRcs.AI

keywords zero-shotcross-domainsequentialrecommendationdomainsemanticbiaslargelanguagemodelsitem-levelgeneralizationpatterntransferinter-domaincompactnessintra-domaindiversityattentionaggregation

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 domain semantic bias, the vocabulary and content-focus differences in LLM-generated item embeddings, is the main obstacle to zero-shot cross-domain sequential recommendation. It proposes LLM-RecG, a model-agnostic framework that trains a generalization loss to align item embeddings across domains while keeping items within each domain distinct, and transfers user behavioral patterns by clustering source sequences and attending to cluster centroids during target inference. The paper reports consistent gains over semantic-only baselines, averaging more than 10%, across multiple base models and domain pairs, including cross-platform transfer. If correct, the framework enables recommendations in unseen domains without target-domain interaction data, using only source interactions and item text.

What carries the argument

The load-bearing object is the generalization loss L_gen (Eq. 13), which combines inter-domain compactness (L_inter, the entropy of each item's similarity to the other domain's center, minimized) with intra-domain diversity (L_intra, the entropy of within-domain similarity, maximized because it enters with a negative coefficient). The item embeddings reshaped by this loss come from an LLM-based semantic encoder followed by a learnable projection layer, and are then consumed by any existing sequential recommender. The sequence-level component is soft sequential pattern attention: k-means centroids of source user sequence embeddings serve as reusable behavior patterns, and a target sequence's cosine similarities to these centroids are softmax-weighted and summed, then concatenated with the target user embedding and projected back. The combination of these two mechanisms is what the paper claims carries the zero-shot transfer.

What would settle it

Measure zero-shot performance after withholding all target-domain item text, validation interactions, and grid-search labels from the training pipeline; if the -RecG gains over -Sem mostly vanish, the reported transfers are not purely zero-shot.

Watch

Extended reading notes

Core claim

LLM-RecG claims that a single training objective can reduce domain semantic bias in LLM-based sequential recommenders and thereby improve zero-shot transfer. At the item level, the generalization loss L_gen = -αL_intra + βL_inter pushes each item's embedding toward the other domain's center while preventing same-domain items from collapsing onto each other, balancing transferability and distinctiveness. At the sequence level, the method clusters source user sequence embeddings with k-means, treats the centroids as transferable behavioral patterns, and computes a soft attention-weighted pattern representation for each target sequence, which is concatenated and projected onto the target user embedding. The paper shows that these additions improve Recall@10 and NDCG@10 over semantic-only baselines for GRU4Rec, SASRec, and BERT4Rec, and that removing intra-domain diversity causes the largest degradation, indicating that preserving fine-grained within-domain distinctions is the most critical part of the mechanism.

Load-bearing premise

The zero-shot claim rests on the premise that using target-domain item text and target-domain validation labels during training and model selection does not count as target-domain training.

Editorial extensions

If this is right

Wrapping LLM-RecG around GRU4Rec, SASRec, or BERT4Rec yields average zero-shot gains exceeding 10% over semantic-only variants, including cross-platform transfers from Steam to Amazon domains.
The framework reduces performance variance across source-target domain pairs: -RecG variants show smaller drops than their -Sem counterparts when the source domain changes.
Intra-domain diversity is the most load-bearing component: ablating it lowers zero-shot performance below the semantic-only baseline, while ablating inter-domain compactness or sequence-level transfer causes smaller drops.
The method also improves in-domain recommendation, suggesting the generalization loss mitigates overfitting to source-domain shortcuts.
A moderate generalization weight (α ≈ 0.001) is necessary; too large a weight degrades performance, so the loss must not overwhelm the ranking objective.

Reading between the lines

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

The protocol uses target-domain item text in the generalization loss and target validation interactions for model selection, so a stricter test that withholds target text would reveal how much of the gain is genuine zero-shot transfer versus target-aware alignment.
Because the method trains only a projection layer and loss weights on top of frozen LLM embeddings, the same objective could be applied to the LLM's own last layer as a cheaper alternative to fine-tuning the entire encoder.
The attention over k-means centroids assumes source and target share coarse behavioral regularities; an adversarial test with domains designed to have opposite temporal patterns would show where the transfer begins to hurt.
The t-SNE evidence suggests the loss trades away some semantic cluster separation for uniformity; a follow-up could quantify the ideal uniformity-alignment tradeoff for each domain pair.

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.

Desk Editor's Note

read the letter

You should know two things about this paper. First, the central inter-domain compactness loss (Eqs. 9–10) is a no-op in the two-domain setting used for every Table 3 row. For any item, the denominator sums over exactly one domain, so Q_id = 1 and L_inter = 0. The ablation in Table 5 shows removing this component changes results, which is a direct contradiction between the equations and the empirical claims. Even if the denominator were corrected to sum over both domain centers, minimizing Q_log Q would drive Q_other toward e^{-1}, not toward 1, so the loss would not implement alignment to the other domain. This is not a typo; the proposed mechanism does not do what the paper says.

Second, the zero-shot characterization is insecure. The generalization loss L_gen is computed over target item embeddings during training (Eqs. 7 and 13), and hyperparameters like alpha are grid-searched using target validation interactions (Section 4.1). The problem definition says D_t includes X_t and forbids training on D_t "directly," so the paper is using target-domain metadata and validation labels to train and select the model. That may be a defensible transductive setting, but it is not the zero-shot setup promised in the abstract.

Give credit where it is due. The paper identifies a real problem—domain semantic bias in LLM embeddings for cross-domain sequential recommendation—and the idea of transferring sequential patterns via k-means clustering plus attention is a reasonable engineering contribution. The experiments are extensive: four domains, three base recommenders, multiple baselines, and code/data are released. The ablation shows that intra-domain diversity and sequence-level generalization both contribute, so those parts of the framework may have value even if the compactness loss is dead as written.

The soft spots are load-bearing. The formal description of the central mechanism is wrong, and the empirical gains cannot be attributed to the stated alignment loss. The zero-shot claim is also undermined by target-data use during training and model selection. These are fixable in a revision, but as submitted the paper's own equations contradict its results.

Who is this for? A reader interested in cross-domain recommendation or in how LLM embeddings can be adapted across domains might find the sequence-level transfer idea worth exploring, but they should not trust the item-level alignment story. It deserves a serious referee because the empirical effort and released code are substantial and the flaws are technical rather than hopeless. My recommendation: reject or major revision in current form, with a request to correct the loss formulation and clarify exactly which target data are used.

Referee Report

3 major / 5 minor

Summary. The paper proposes LLM-RecG, a model-agnostic framework for zero-shot cross-domain sequential recommendation (ZCDSR). It uses LLM-generated item embeddings projected into a latent space, trains a sequential recommender on a source domain, and adds two item-level generalization terms—inter-domain compactness (L_inter) and intra-domain diversity (L_intra)—plus a sequence-level transfer module that clusters source user sequence embeddings and applies attention over cluster centroids at target inference. Experiments on Amazon subsets and Steam report consistent gains over semantic-only baselines across GRU4Rec, SASRec, and BERT4Rec, with ablations, sensitivity analysis, and t-SNE visualizations.

Significance. If the inter-domain compactness mechanism were active, the framework would be a plausible contribution to zero-shot sequential recommendation, and the breadth of architectures and datasets is a strength. However, the central item-level alignment term is arithmetically zero in the two-domain setting used in every experiment, so the proposed mechanism is not present in the published formulation; the sequence-level component alone cannot support the paper's central claims as stated. The paper also leaves the zero-shot protocol ambiguous with respect to target-domain metadata and validation. As a result, the contribution is not established.

major comments (3)

[Section 3.2, Eqs. (9)-(10)] L_inter is identically zero in the two-domain configuration used for every row of Table 3. For an item in domain d_i, the sum over d in {D_s,D_t} with d != d_i has exactly one term, and the denominator of Q_id in Eq. (10) also has exactly one term (the other domain center), so Q_id = exp(x)/exp(x) = 1 and every addend is 1 * log 1 = 0. Consequently L_inter has no gradient and cannot align embeddings. This contradicts the abstract and Section 3.2's claim that inter-domain compactness is the route to cross-domain alignment, and it is inconsistent with Table 5, where removing IC changes scores; the paper must either correct the loss definition or explain the ablation under the published equations.
[Section 3.2, Eqs. (9)-(10)] Even a corrected softmax over both domain centers would not implement the stated alignment: minimizing the two-class quantity Q_other log Q_other drives Q_other toward e^{-1} approximately 0.37, not toward 1, so the proposed objective does not encourage membership in the other domain. The mechanism described in Definition 3.1 therefore requires a different formulation.
[Sections 3.2, 4.1.1, 4.1.3] The zero-shot protocol is not cleanly specified. L_gen in Eq. (13) is computed over both source and target item embeddings via Eq. (7), so target-domain item text enters training; the penultimate target interaction is held out for validation (Section 4.1.1), and hyperparameters including alpha are selected by grid search. The paper never states which target data (item metadata, validation interactions, or neither) are permitted, and if target validation labels influence model selection, the reported numbers are not purely zero-shot. This should be clarified and, if target validation is used, the claims should be adjusted.

minor comments (5)

[Section 3.1] Typo: 'exisiting' should be 'existing'; in Section 4.2.1, 'transfering' should be 'transferring'.
[Section 3] The acronym 'ZSCDSR' is used once after 'ZCDSR' is defined; use the acronym consistently.
[Section 4.1.3] The alpha search range '0.05, 0.01,, 0.005' contains a doubled comma.
[Section 4.2.1] The sentence 'BERT4Rec-RecG by relatively 28.9%' is incomplete; it should state which metric and which baseline the improvement refers to.
[Section 3.2, Eq. (14)] Equation (14) is typeset ambiguously; the intended scaling, likely beta = alpha * |N| / |{D_s,D_t}|^3, should be written explicitly.

Circularity Check

1 steps flagged · score 6.0 of 10

Inter-domain compactness loss is identically zero as written, reducing the paper's central alignment claim to a no-op by construction.

other [Section 3.2.1, Eqs. (9) and (10)]
"Linter = Σ_{v_i∈V} Σ_{d∈{D_s,D_t}, d≠d_i} Q_id log Q_id, (9) ... Q_id = exp(cos(e_i^d,c_d)/τ) / Σ_{d'∈{D_s,D_t}, d'≠d_i} exp(cos(e_i^d,c_{d'})/τ), (10) ... Minimizing L_inter encourages embeddings to align with the centers of other domains, enhancing inter-domain compactness and reducing domain semantic biases."

In every experiment reported in Table 3 there are exactly two domains, so for any item with domain d_i, the condition d'≠d_i leaves a single domain d'=d in the denominator of Eq. (10). The denominator is therefore term-for-term identical to the numerator, giving Q_id=1 for every item and every d≠d_i. Each addend of Eq. (9) is 1·log1=0, so L_inter≡0 and has no gradient. The claimed inter-domain alignment mechanism is thus a no-op by construction under the published equations. Table 5's ablation, which attributes a performance drop to removing IC, cannot be explained by Eq. (9)-(10); either the released code uses a different normalization (making the paper's formal description wrong) or the formal loss is vacuous.

full rationale

The empirical evaluation is self-contained: performance is measured against external baselines (GRU4Rec, SASRec, BERT4Rec, UniSRec, RecFormer) on standard Amazon and Steam datasets, and the code is released, so the results are not circular in the usual fitted-prediction sense. The self-citations [37,38] only motivate the existence of cross-domain behavioral similarity and are not load-bearing for the method's derivation. However, one central formal step is a by-construction no-op: the inter-domain compactness term, which the abstract and Section 3.2 present as the route to cross-domain alignment, is identically zero under Eqs. (9)-(10) in the two-domain setting used throughout. This makes the claimed alignment mechanism and the IC ablation in Table 5 formally impossible as written. A separate correctness concern, distinct from circularity, is that target-domain item text and target validation labels enter training and model selection (Eqs. (7) and (13), Section 4.1.1 penultimate-item validation, Section 4.1.3 grid search over alpha), which undermines the zero-shot characterization but does not by itself raise the circularity score. The score of 6 reflects the partial circular reduction of the paper's central item-level mechanism by construction, while the remaining framework (intra-domain diversity, sequential pattern attention) still has independent content.

Assumptions & free parameters 5 free parameters · 3 assumptions · 0 invented entities

The framework relies on several tuned hyperparameters, two domain assumptions about cross-domain behavioral similarity and target-data usage, and standard mathematical primitives. No new physical or conceptual entities are introduced; sequential pattern centroids are learned model parameters, not postulated entities.

free parameters (5)

generalization weight alpha = 0.001 (optimal in sensitivity analysis)
Balances L_rec and L_gen; selected by grid search per model, not derived from theory.
inter-domain weight beta = alpha * |N| / |{D_s,D_t}|^3 (Eq. 14)
Set by a formula from alpha and domain/item counts; no independent justification for the scaling.
temperature tau = not reported
Controls sharpness of similarity distributions in Eqs. 10 and 12; values are not stated in the paper.
number of sequential patterns k = not reported
k-means cluster count in Eq. 16; the paper does not report the value or its sensitivity.
projection dimension d_l = searched over {64, 128, 256}
Embedding dimensionality selected by grid search; affects all downstream losses.

assumptions (3)

domain assumption Source-domain user behavioral patterns, captured as cluster centroids, transfer to unseen target domains.
Section 3.3 uses k-means centroids of source sequence embeddings as priors for target sequences; this presumes cross-domain similarity of sequential behavior, supported only by citations [28,38].
ad hoc to paper Target-domain item metadata may be used in training and target validation may be used for model selection without invalidating zero-shot claims.
Section 3.2 computes L_gen using target domain centers and embeddings; Section 4.1.1 defines target validation splits and Section 4.1.3 tunes hyperparameters. The paper does not justify this against its own zero-shot definition in Section 2.
standard math BPR ranking loss, k-means clustering, and softmax attention are valid optimization and inference primitives.
Used in Eqs. 6, 16, and 18; these are standard tools and not contested.

how reviews work

0 comments

Cite this review

Pith. "Pith review of LLM-RecG: A Semantic Bias-Aware Framework for Zero-Shot Sequential Recommendation." pith.science (2026). https://pith.science/paper/7EH4EUVC

@misc{pith2026250119232,
  author       = {Pith},
  title        = {Pith review of: LLM-RecG: A Semantic Bias-Aware Framework for Zero-Shot Sequential Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7EH4EUVC}},
  note         = {Machine review of arXiv:2501.19232}
}

read the original abstract

Zero-shot cross-domain sequential recommendation (ZCDSR) enables predictions in unseen domains without additional training or fine-tuning, addressing the limitations of traditional models in sparse data environments. Recent advancements in large language models (LLMs) have significantly enhanced ZCDSR by facilitating cross-domain knowledge transfer through rich, pretrained representations. Despite this progress, domain semantic bias -- arising from differences in vocabulary and content focus between domains -- remains a persistent challenge, leading to misaligned item embeddings and reduced generalization across domains. To address this, we propose a novel semantic bias-aware framework that enhances LLM-based ZCDSR by improving cross-domain alignment at both the item and sequential levels. At the item level, we introduce a generalization loss that aligns the embeddings of items across domains (inter-domain compactness), while preserving the unique characteristics of each item within its own domain (intra-domain diversity). This ensures that item embeddings can be transferred effectively between domains without collapsing into overly generic or uniform representations. At the sequential level, we develop a method to transfer user behavioral patterns by clustering source domain user sequences and applying attention-based aggregation during target domain inference. We dynamically adapt user embeddings to unseen domains, enabling effective zero-shot recommendations without requiring target-domain interactions...

Figures

Figures reproduced from arXiv: 2501.19232 by the authors.

**Figure 1.** The model framework of LLM-RecG. between positive and negative items. For a given user 𝑢 𝑠 𝑗 in the source domain, the loss is defined as: Lrec = − ∑︁ 𝑗 ∑︁ (𝑖 +,𝑖− ) log 𝜎 score(𝑢 𝑠 𝑗 , 𝑣𝑠 𝑖 + ) − score(𝑢 𝑠 𝑗 , 𝑣𝑠 𝑖 − ) , (6) where 𝑖 + and 𝑖 − are positive and negative items, respectively, and 𝜎(·) is the sigmoid function. When performing zero-shot inference on a new domain, the same pretrained E and projection … view at source ↗

**Figure 3.** Visualization of item embeddings on the Video Games do [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

**Figure 2.** Impact of the generalization weight 𝛼 on ZCDSR performance over IS (left) and MI (right) datasets. A moderate value of 𝛼 = 0.001 consistently yields the best performance, effectively balancing recommendation and generalization loss. Larger 𝛼 values lead to performance degradation, highlighting the importance of tuning this parameter. 4.4 Sensitivity Analysis(RQ3) To evaluate the impacts of key parameter 𝛼 in ZCDSR… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

SelfElicit: Your Language Model Secretly Knows Where is the Relevant Evidence
cs.CL 2025-02 conditional novelty 6.0 of 10

SelfElicit uses deep-layer attention to automatically highlight relevant evidence sentences in the input context, yielding consistent QA accuracy gains across six instruction-tuned LLMs.
Large Language Model Enhanced Recommender Systems: A Survey
cs.IR 2024-12 unverdicted novelty 4.0 of 10

A survey organizing LLM-enhanced recommender systems into knowledge, interaction, and model enhancement, and tracing a shift from explicit text to implicit embeddings and fine-tuned open-source LLMs.

Reference graph

Works this paper leans on

52 extracted references · 20 canonical work pages · cited by 2 Pith papers

[1]

Keqin Bao, Jizhi Zhang, Yang Zhang, Wenjie Wang, Fuli Feng, and Xiangnan He. 2023. Tallrec: An effective and efficient tuning framework to align large language model with recommendation. InProceedings of the 17th ACM Conference on Recommender Systems. 1007–1014

2023
[2]

Parishad BehnamGhader, Vaibhav Adlakha, Marius Mosbach, Dzmitry Bahdanau, Nicolas Chapados, and Siva Reddy. 2024. LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders. In First Conference on Language Modeling . https://openreview.net/forum?id=IW1PR7vEBf

work page 2024
[3]

Jiangxia Cao, Xin Cong, Jiawei Sheng, Tingwen Liu, and Bin Wang. 2022. Con- trastive cross-domain sequential recommendation. InProceedings of the 31st ACM International Conference on Information & Knowledge Management . 138–147

work page 2022
[4]

Jacob Devlin. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

arXiv 2018
[5]

Hao Ding, Yifei Ma, Anoop Deoras, Yuyang Wang, and Hao Wang. 2021. Zero- shot recommender systems. arXiv preprint arXiv:2105.08318 (2021)

arXiv 2021
[6]

Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783 (2024)

arXiv 2024
[7]

Philip J Feng, Pingjun Pan, Tingting Zhou, Hongxiang Chen, and Chuanjiang Luo. 2021. Zero shot on the cold-start problem: Model-agnostic interest learning for recommender systems. InProceedings of the 30th ACM international conference on information & knowledge management . 474–483

work page 2021
[8]

Zichuan Fu, Xiangyang Li, Chuhan Wu, Yichao Wang, Kuicai Dong, Xiangyu Zhao, Mengchen Zhao, Huifeng Guo, and Ruiming Tang. 2023. A unified frame- work for multi-domain ctr prediction via large language models. ACM Transac- tions on Information Systems (2023)

2023

Show all 52 references

[9]

Yunfan Gao, Tao Sheng, Youlin Xiang, Yun Xiong, Haofen Wang, and Jiawei Zhang. 2023. Chat-rec: Towards interactive and explainable llms-augmented recommender system. arXiv preprint arXiv:2303.14524 (2023)

2023 arXiv
[10]

Jesse Harte, Wouter Zorgdrager, Panos Louridas, Asterios Katsifodimos, Diet- mar Jannach, and Marios Fragkoulis. 2023. Leveraging large language models for sequential recommendation. In Proceedings of the 17th ACM Conference on Recommender Systems. 1096–1102

2023
[11]

Ruining He and Julian McAuley. 2016. Fusing similarity models with markov chains for sparse sequential recommendation. In 2016 IEEE 16th international conference on data mining (ICDM) . IEEE, 191–200

2016
[12]

Zhankui He, Zhouhang Xie, Rahul Jha, Harald Steck, Dawen Liang, Yesu Feng, Bodhisattwa Prasad Majumder, Nathan Kallus, and Julian McAuley. 2023. Large language models as zero-shot conversational recommenders. InProceedings of the 32nd ACM international conference on informatio...

2023
[13]

B Hidasi. 2015. Session-based Recommendations with Recurrent Neural Networks. arXiv preprint arXiv:1511.06939 (2015)

2015 arXiv
[14]

Yupeng Hou, Jiacheng Li, Zhankui He, An Yan, Xiusi Chen, and Julian McAuley
[15]

Yupeng Hou, Shanlei Mu, Wayne Xin Zhao, Yaliang Li, Bolin Ding, and Ji-Rong Wen. 2022. Towards universal sequence representation learning for recommender systems. In Proceedings of the 28th ACM SIGKDD Conference on Knowledge Dis- covery and Data Mining . 585–593

2022
[16]

Jianchao Ji, Zelong Li, Shuyuan Xu, Wenyue Hua, Yingqiang Ge, Juntao Tan, and Yongfeng Zhang. 2024. Genrec: Large language model for generative recommen- dation. In European Conference on Information Retrieval . Springer, 494–502

2024
[17]

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

2018
[18]

Jiacheng Li, Ming Wang, Jin Li, Jinmiao Fu, Xin Shen, Jingbo Shang, and Julian McAuley. 2023. Text is all you need: Learning language representations for sequential recommendation. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . 1258–1267

2023
[19]

Ruyu Li, Wenhao Deng, Yu Cheng, Zheng Yuan, Jiaqi Zhang, and Fajie Yuan
[20]

Yunzhe Li, Yue Ding, Bo Chen, Xin Xin, Yule Wang, Yuxiang Shi, Ruiming Tang, and Dong Wang. 2021. Extracting Attentive Social Temporal Excitation for Sequential Recommendation. In Proceedings of the 30th ACM International Con- ference on Information & Knowledge Management . 998–1007

2021
[21]

Youhua Li, Hanwen Du, Yongxin Ni, Pengpeng Zhao, Qi Guo, Fajie Yuan, and Xi- aofang Zhou. 2024. Multi-modality is all you need for transferable recommender systems. In 2024 IEEE 40th International Conference on Data Engineering (ICDE) . IEEE, 5008–5021

2024
[22]

Qijiong Liu, Nuo Chen, Tetsuya Sakai, and Xiao-Ming Wu. 2024. Once: Boosting content-based recommendation with both open-and closed-source large language models. In Proceedings of the 17th ACM International Conference on Web Search and Data Mining. 452–461. RecSys ’25, Septemb...

2024
[23]

Qidong Liu, Xian Wu, Yejing Wang, Zijian Zhang, Feng Tian, Yefeng Zheng, and Xiangyu Zhao. [n. d.]. LLM-ESR: Large Language Models Enhancement for Long-tailed Sequential Recommendation. In The Thirty-eighth Annual Conference on Neural Information Processing Systems
[24]

Zhongzhou Liu, Hao Zhang, Kuicai Dong, and Yuan Fang. 2024. Collaborative Cross-modal Fusion with Large Language Model for Recommendation. In Pro- ceedings of the 33rd ACM International Conference on Information and Knowledge Management. 1565–1574

2024
[25]

Linhao Luo, Yumeng Li, Buyu Gao, Shuai Tang, Sinan Wang, Jiancheng Li, Tan- chao Zhu, Jiancai Liu, Zhao Li, and Shirui Pan. 2023. MAMDR: A model agnostic learning framework for multi-domain recommendation. In 2023 IEEE 39th Inter- national Conference on Data Engineering (ICDE)...

2023
[26]

2019.𝜋-net: A parallel information-sharing network for shared-account cross- domain sequential recommendations

Muyang Ma, Pengjie Ren, Yujie Lin, Zhumin Chen, Jun Ma, and Maarten de Rijke. 2019.𝜋-net: A parallel information-sharing network for shared-account cross- domain sequential recommendations. In Proceedings of the 42nd international ACM SIGIR conference on research and developme...

2019
[27]

James MacQueen et al. 1967. Some methods for classification and analysis of multivariate observations. In Proceedings of the fifth Berkeley symposium on mathematical statistics and probability , Vol. 1. Oakland, CA, USA, 281–297

1967
[28]

Vilfredo Pareto. 1897. Cursus de l’économie politique . F. Rouge, Lausanne
[29]

Zhaopeng Qiu, Xian Wu, Jingyue Gao, and Wei Fan. 2021. U-BERT: Pre-training user representations for improved recommendation. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 4320–4327

2021
[30]

Shashank Rajput, Nikhil Mehta, Anima Singh, Raghunandan Hulikal Keshavan, Trung Vu, Lukasz Heldt, Lichan Hong, Yi Tay, Vinh Tran, Jonah Samost, et al
[31]

Steffen Rendle, Christoph Freudenthaler, Zeno Gantner, and Lars Schmidt-Thieme
[32]

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

2010
[33]

Advances in Neural Information Processing Systems 36 (2023), 10299–10315

Recommender systems with generative retrieval. Advances in Neural Information Processing Systems 36 (2023), 10299–10315

2023
[34]

Fei Sun, Jun Liu, Jian Wu, Changhua Pei, Xiao Lin, Wenwu Ou, and Peng Jiang
[35]

Wenchao Sun, Muyang Ma, Pengjie Ren, Yujie Lin, Zhumin Chen, Zhaochun Ren, Jun Ma, and Maarten De Rijke. 2021. Parallel split-join networks for shared account cross-domain sequential recommendations. IEEE Transactions on Knowl- edge and Data Engineering 35, 4 (2021), 4106–4123

2021
[36]

Rohan Taori, Ishaan Gulrajani, Tianyi Zhang, Yann Dubois, Xuechen Li, Carlos Guestrin, Percy Liang, and Tatsunori B Hashimoto. 2023. Stanford alpaca: An instruction-following llama model

2023
[37]

Xiang-Rong Sheng, Liqin Zhao, Guorui Zhou, Xinyao Ding, Binding Dai, Qiang Luo, Siran Yang, Jingshan Lv, Chi Zhang, Hongbo Deng, et al. 2021. One model to serve all: Star topology adaptive recommender for multi-domain ctr prediction. In Proceedings of the 30th ACM Internationa...

2021
[38]

Junting Wang, Praneet Rathi, and Hari Sundaram. 2024. A Pre-trained Zero-shot Sequential Recommendation Framework via Popularity Dynamics. In 18th ACM Conference on Recommender Systems (RecSys ’24) . ACM, 433–443. doi:10.1145/ 3640457.3688145

2024
[39]

Yaqing Wang, Caili Guo, Yunfei Chu, Jenq-Neng Hwang, and Chunyan Feng
[40]

Yuhao Wang, Yichao Wang, Zichuan Fu, Xiangyang Li, Wanyu Wang, Yuyang Ye, Xiangyu Zhao, Huifeng Guo, and Ruiming Tang. 2024. Llm4msr: An llm- enhanced paradigm for multi-scenario recommendation. In Proceedings of the 33rd ACM International Conference on Information and Knowled...

2024
[41]

Yule Wang, Xin Xin, Yue Ding, Yunzhe Li, and Dong Wang. 2021. ICPE: An item cluster-wise pareto-efficient framework for recommendation debiasing. arXiv preprint arXiv:2109.12887 (2021)

2021 arXiv
[42]

Junting Wang, Adit Krishnan, Hari Sundaram, and Yunzhe Li. 2023. Pre-trained Neural Recommenders: A Transferable Zero-Shot Framework for Recommenda- tion Systems. arXiv preprint arXiv:2309.01188 (2023)

2023 arXiv
[43]

Wujiang Xu, Qitian Wu, Runzhong Wang, Mingming Ha, Qiongxu Ma, Linxun Chen, Bing Han, and Junchi Yan. 2024. Rethinking cross-domain sequential recommendation under open-world assumptions. In Proceedings of the ACM on Web Conference 2024. 3173–3184

2024
[44]

Guang Yang, Xiaoguang Hong, Zhaohui Peng, and Yang Xu. 2020. Long Short- Term Memory with Sequence Completion for Cross-Domain Sequential Rec- ommendation. In APWeb/W AIM. https://api.semanticscholar.org/CorpusID: 224770681

2020
[45]

Zhengyi Yang, Jiancan Wu, Yanchen Luo, Jizhi Zhang, Yancheng Yuan, An Zhang, Xiang Wang, and Xiangnan He. 2023. Large language model can interpret latent space of sequential recommender. arXiv preprint arXiv:2310.20487 (2023)

2023 arXiv
[46]

Zhenrui Yue, Sara Rabhi, Gabriel de Souza Pereira Moreira, Dong Wang, and Even Oldridge. 2023. LlamaRec: Two-stage recommendation using large language models for ranking. arXiv preprint arXiv:2311.02089 (2023)

2023 arXiv
[48]

Likang Wu, Zhi Zheng, Zhaopeng Qiu, Hao Wang, Hongchao Gu, Tingjia Shen, Chuan Qin, Chen Zhu, Hengshu Zhu, Qi Liu, et al . 2024. A survey on large language models for recommendation. World Wide Web 27, 5 (2024), 60

2024
[2012]

arXiv preprint arXiv:1205.2618 (2012)

BPR: Bayesian personalized ranking from implicit feedback. arXiv preprint arXiv:1205.2618 (2012)

2012 arXiv
[2019]

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

BERT4Rec: Sequential recommendation with bidirectional encoder rep- resentations from transformer. In Proceedings of the 28th ACM international conference on information and knowledge management . 1441–1450
[2020]

Neurocomputing 380 (2020), 271–284

A cross-domain hierarchical recurrent model for personalized session- based recommendations. Neurocomputing 380 (2020), 271–284. https://api. semanticscholar.org/CorpusID:209903109

2020
[2023]

Exploring the upper limits of text-based collaborative filtering using large language models: Discoveries and insights.arXiv preprint arXiv:2305.11700 (2023)

2023
[2024]

arXiv preprint arXiv:2403.03952 (2024)

Bridging Language and Items for Retrieval and Recommendation. arXiv preprint arXiv:2403.03952 (2024)

2024 arXiv

Pith tools

Reviewed August 9, 2026 · model on record in the stance chip above.

[1] [1]

Keqin Bao, Jizhi Zhang, Yang Zhang, Wenjie Wang, Fuli Feng, and Xiangnan He. 2023. Tallrec: An effective and efficient tuning framework to align large language model with recommendation. InProceedings of the 17th ACM Conference on Recommender Systems. 1007–1014

2023

[2] [2]

Parishad BehnamGhader, Vaibhav Adlakha, Marius Mosbach, Dzmitry Bahdanau, Nicolas Chapados, and Siva Reddy. 2024. LLM2Vec: Large Language Models Are Secretly Powerful Text Encoders. In First Conference on Language Modeling . https://openreview.net/forum?id=IW1PR7vEBf

work page 2024

[3] [3]

Jiangxia Cao, Xin Cong, Jiawei Sheng, Tingwen Liu, and Bin Wang. 2022. Con- trastive cross-domain sequential recommendation. InProceedings of the 31st ACM International Conference on Information & Knowledge Management . 138–147

work page 2022

[4] [4]

Jacob Devlin. 2018. Bert: Pre-training of deep bidirectional transformers for language understanding. arXiv preprint arXiv:1810.04805 (2018)

arXiv 2018

[5] [5]

Hao Ding, Yifei Ma, Anoop Deoras, Yuyang Wang, and Hao Wang. 2021. Zero- shot recommender systems. arXiv preprint arXiv:2105.08318 (2021)

arXiv 2021

[6] [6]

Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Amy Yang, Angela Fan, et al. 2024. The llama 3 herd of models. arXiv preprint arXiv:2407.21783 (2024)

arXiv 2024

[7] [7]

Philip J Feng, Pingjun Pan, Tingting Zhou, Hongxiang Chen, and Chuanjiang Luo. 2021. Zero shot on the cold-start problem: Model-agnostic interest learning for recommender systems. InProceedings of the 30th ACM international conference on information & knowledge management . 474–483

work page 2021

[8] [8]

Zichuan Fu, Xiangyang Li, Chuhan Wu, Yichao Wang, Kuicai Dong, Xiangyu Zhao, Mengchen Zhao, Huifeng Guo, and Ruiming Tang. 2023. A unified frame- work for multi-domain ctr prediction via large language models. ACM Transac- tions on Information Systems (2023)

2023