pith. sign in

arxiv: 2605.00499 · v1 · submitted 2026-05-01 · 💻 cs.IR

Time-Interval-Aware Disentangled Expert Modeling for Next-Basket Recommendation

Zhiying Deng , Yuan Fu , Usman Farooq , Ziwei Tian , Wei Liu , Jianjun Li This is my paper

Pith reviewed 2026-05-09 18:54 UTC · model grok-4.3

classification 💻 cs.IR
keywords next-basket recommendationdisentangled expertstime-interval modelinghabitual repurchaseexploratory interestdual-expert architecturerecommendation systems
0
0 comments X

The pith

TIDE improves next-basket prediction by using dual experts to separate habitual repurchase from exploratory discovery while modeling time intervals.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper establishes that next-basket recommendation suffers when models mix a user's repeating purchase habits with their desire to try new items in one representation and when they treat time as discrete steps rather than continuous intervals with item-specific patterns. TIDE counters this by introducing a dual-expert setup where one expert handles recurring needs and the other guides discovery, controlled by an item-aware gate, plus a time encoding that captures periodicities and decay. A sympathetic reader would care because successful separation should produce baskets that respect both stability and novelty without one dominating the other. The experiments show this yields higher accuracy than prior methods across four datasets.

Core claim

TIDE addresses entangled intents and discrete time modeling in NBR by combining a Hawkes-enhanced Fourier Time Encoding to capture item-specific temporal periodicities and dynamic decay with a dual-expert architecture: a Habit Expert for recurring needs and a Pattern-Guided Exploration Expert for discovery, integrated via an item-aware gating mechanism that adaptively balances the two.

What carries the argument

Dual-expert architecture with Habit Expert and Pattern-Guided Exploration Expert, controlled by item-aware gating and paired with Hawkes-enhanced Fourier Time Encoding for continuous time modeling.

If this is right

  • TIDE outperforms representative state-of-the-art NBR methods on four diverse real-world datasets.
  • The item-aware gate lets the model shift emphasis between experts depending on the specific item context.
  • Item-specific periodicities and dynamic decay are explicitly modeled instead of being ignored by discrete sequences.
  • Habitual and exploratory motives no longer compete inside a single shared representation.

Where Pith is reading between the lines

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

  • The same separation of stable versus novel intent could be tested in session-based recommendation outside baskets.
  • The time encoding component might transfer to other sequential tasks where item periodicity varies, such as content consumption.
  • Ablation studies isolating each expert could reveal whether one intent type dominates on particular user groups.
  • If the gating works as described, the model outputs could be made more interpretable by surfacing which expert drove each recommended item.

Load-bearing premise

The dual-expert structure with item-aware gating can cleanly separate habitual repurchase from exploratory interest without discarding useful interactions or adding new biases that hurt overall accuracy.

What would settle it

Run TIDE and the strongest baseline on a new dataset where repurchase habits and new-item trials are tightly coupled in the same sessions; if TIDE no longer outperforms or if removing the gating or time encoding leaves performance unchanged, the separation claim does not hold.

Figures

Figures reproduced from arXiv: 2605.00499 by Jianjun Li, Usman Farooq, Wei Liu, Yuan Fu, Zhiying Deng, Ziwei Tian.

Figure 1
Figure 1. Figure 1: Motivation of TIDE. (a) The NBR task aims to predict view at source ↗
Figure 2
Figure 2. Figure 2: Statistical analysis of user consumption behaviors. (a) Evolution of Repeat Ratio. (b) Distribution of Repurchase view at source ↗
Figure 3
Figure 3. Figure 3: TIDE architecture, which consists of three key modules: (1) Habit Expert, which captures non-monotonic replenishment view at source ↗
Figure 4
Figure 4. Figure 4: Probability density distribution of semantic align view at source ↗
Figure 5
Figure 5. Figure 5: t-SNE visualization of the evolution of habit intent ( view at source ↗
Figure 6
Figure 6. Figure 6: examines the impact of the contrastive loss weight 𝜇 and the minimum support count on TIDE’s performance. 1) Contrastive Loss Weight 𝜇: Performance across all datasets follows an inverted￾U trend. This suggests that while moderate alignment is vital for bridging the semantic gap between experts, excessive weight over￾prioritizes proximity at the expense of the model’s discriminative power. 2) Minimum Suppo… view at source ↗
read the original abstract

Next-basket recommendation (NBR) is a type of recommendation that aims to predict a set of items a user will purchase based on their historical transaction basket sequences. It is governed by a dynamic interplay between two distinct user intents: habitual repurchase, which involves repeating past behaviors, and exploratory interest, which involves discovering new items. However, existing NBR methods generally suffer from two limitations: (1) they often entangle these conflicting motives within a single representation, causing habits to overshadow discovery, and (2) they rely on discrete sequential modeling that ignores continuous-time intervals and item-specific periodicities. In this paper, we propose a novel solution named Time-Interval Disentangled Experts (TIDE) to address these challenges. TIDE incorporates a Hawkes-enhanced Fourier Time Encoding to capture item-specific temporal periodicities and dynamic decay. To decouple user intentions, TIDE utilizes a dual-expert architecture that integrates a Habit Expert for recurring needs and a Pattern-Guided Exploration Expert for discovery. Combined with an item-aware gating mechanism, TIDE adaptively balances repurchase and exploration. Extensive experiments on four diverse real-world datasets demonstrate that TIDE consistently outperforms representative state-of-the-art NBR methods.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The manuscript proposes TIDE for next-basket recommendation, which models the interplay between habitual repurchase and exploratory interest via a dual-expert architecture (Habit Expert plus Pattern-Guided Exploration Expert) and an item-aware gating mechanism. It augments this with a Hawkes-enhanced Fourier Time Encoding to capture item-specific periodicities and continuous-time decay, claiming that the resulting disentanglement yields consistent outperformance over representative state-of-the-art NBR baselines on four real-world datasets.

Significance. If the disentanglement claim is substantiated, the work would address a recognized limitation in sequential recommendation by explicitly separating conflicting user intents rather than relying on a single entangled representation. The continuous-time component also fills a gap left by discrete basket-sequence models. Credit is due for the architectural integration of Hawkes processes with Fourier encodings and the explicit dual-expert design; however, the significance hinges on whether observed gains arise from successful separation rather than added capacity.

major comments (3)
  1. [§4] §4 (Experiments): no ablation holds total parameter count fixed when comparing the dual-expert model against a single-expert baseline of matched capacity. Without this control, it remains unclear whether reported gains on the four datasets are driven by the disentanglement mechanism or simply by increased expressivity.
  2. [§3.3] §3.3 (Item-aware Gating): the gating equations are defined, yet the manuscript supplies no post-hoc diagnostics such as expert activation histograms, correlation between the two expert outputs, or intent-specific recall breakdowns. These diagnostics are load-bearing for the central claim that the architecture successfully decouples repurchase from exploration.
  3. [§4.2] §4.2 (Results): the tables report point estimates without statistical significance tests (e.g., paired t-tests across multiple random seeds) or standard deviations. This weakens the assertion that TIDE “consistently outperforms” the baselines.
minor comments (2)
  1. [§3.1] Notation for the Hawkes intensity function and Fourier coefficients is introduced without an explicit reference to the original Hawkes process formulation; a brief citation would improve clarity.
  2. [Figure 2] Figure 2 (model overview) would benefit from an explicit legend distinguishing the two experts and the gating path.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thoughtful and constructive comments. We address each major comment below and will incorporate revisions to strengthen the manuscript.

read point-by-point responses

  1. Referee: [§4] §4 (Experiments): no ablation holds total parameter count fixed when comparing the dual-expert model against a single-expert baseline of matched capacity. Without this control, it remains unclear whether reported gains on the four datasets are driven by the disentanglement mechanism or simply by increased expressivity.

    Authors: We agree that controlling for model capacity is necessary to isolate the contribution of the dual-expert disentanglement. In the revised manuscript we will add an ablation in which the single-expert baseline is enlarged (by increasing hidden dimension and/or number of layers) until its total parameter count matches that of TIDE. Results on all four datasets will be reported alongside the original single-expert baseline. revision: yes

  2. Referee: [§3.3] §3.3 (Item-aware Gating): the gating equations are defined, yet the manuscript supplies no post-hoc diagnostics such as expert activation histograms, correlation between the two expert outputs, or intent-specific recall breakdowns. These diagnostics are load-bearing for the central claim that the architecture successfully decouples repurchase from exploration.

    Authors: We acknowledge that direct evidence of successful disentanglement is currently missing. We will add three diagnostics to the revised Section 3.3 and/or a new experimental subsection: (1) histograms of the item-aware gating weights across users and items, (2) Pearson correlations between the output embeddings of the Habit Expert and the Pattern-Guided Exploration Expert, and (3) intent-specific recall@K obtained by partitioning test items into habitual (frequent repurchase) and exploratory (first-time or rare) categories. These analyses will be performed on the same four datasets. revision: yes

  3. Referee: [§4.2] §4.2 (Results): the tables report point estimates without statistical significance tests (e.g., paired t-tests across multiple random seeds) or standard deviations. This weakens the assertion that TIDE “consistently outperforms” the baselines.

    Authors: We agree that reporting only point estimates limits the strength of the performance claims. In the revision we will rerun all experiments with five different random seeds, report mean and standard deviation for every metric, and include paired t-test p-values comparing TIDE against each baseline. Updated tables will appear in Section 4.2. revision: yes

Circularity Check

0 steps flagged

No circularity: architectural claims and empirical results are independent of self-referential reductions.

full rationale

The paper presents TIDE as a novel integration of Hawkes-enhanced Fourier time encoding, a dual-expert (Habit + Pattern-Guided Exploration) architecture, and item-aware gating to address entanglement and discrete-time limitations in NBR. No equations, predictions, or uniqueness claims in the abstract or described components reduce by construction to fitted inputs, self-citations, or renamed known results. Outperformance is asserted via experiments on four datasets rather than tautological redefinitions, leaving the derivation self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 0 axioms · 0 invented entities

The central claim depends on several learned neural-network parameters whose exact count and initialization are not specified in the abstract; the time-encoding and gating mechanisms introduce additional tunable components whose values are fitted to training data. No explicit axioms or invented physical entities are stated.

free parameters (2)
  • neural network weights and biases
    All parameters in the dual experts, time encoder, and gating network are learned from data during training.
  • Hawkes process and Fourier coefficients
    Parameters controlling decay rates and periodic components in the time encoding are fitted to observed purchase intervals.

pith-pipeline@v0.9.0 · 5524 in / 1224 out tokens · 41656 ms · 2026-05-09T18:54:51.347837+00:00 · methodology

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

52 extracted references

  1. [1]

    Mozhdeh Ariannezhad, Sami Jullien, Ming Li, Min Fang, Sebastian Schelter, and Maarten de Rijke. 2022. ReCANet: A Repeat Consumption-Aware Neural Network for Next Basket Recommendation in Grocery Shopping. InSIGIR ’22: The 45th International ACM SIGIR Conference on Research and Development in Information Retrieval. 1240–1250

    2022

  2. [2]

    Ilwoong Baek, Mincheol Yoon, Seongmin Park, and Jongwuk Lee. 2025. Muffin: Mixture of user-adaptive frequency filtering for sequential recommendation. InProceedings of the 34th ACM International Conference on Information and Knowledge Management. 119–128

    2025

  3. [3]

    Ruichu Cai, Siyu Wu, Jie Qiao, Zhifeng Hao, Keli Zhang, and Xi Zhang. 2024. THPs: Topological Hawkes Processes for Learning Causal Structure on Event Sequences.IEEE Transactions on Neural Networks and Learning Systems35, 1 (2024), 479–493

    2024

  4. [4]

    Yizhou Dang, Enneng Yang, Guibing Guo, Linying Jiang, Xingwei Wang, Xiaoxiao Xu, Qinghui Sun, and Hong Liu. 2023. Uniform sequence better: Time interval aware data augmentation for sequential recommendation. InProceedings of the AAAI conference on artificial intelligence, Vol. 37. 4225–4232

    2023

  5. [5]

    Zhiying Deng, Jianjun Li, Zhiqiang Guo, Wei Liu, Li Zou, and Guohui Li. 2023. Multi-view multi-aspect neural networks for next-basket recommendation. In Proceedings of the 46th international ACM SIGIR conference on research and devel- opment in information retrieval. 1283–1292

    2023

  6. [6]

    Zhiying Deng, Jianjun Li, Wei Liu, and Juan Zhao. 2025. Unbiased Interest Modeling in Sequential Basket Analysis: Addressing Repetition Bias with Multi- Factor Estimation.Transactions on Recommender Systems3, 4 (2025), 54:1–54:27

    2025

  7. [7]

    Zhiying Deng, Jianjun Li, Li Zou, Wei Liu, Si Shi, Qian Chen, Juan Zhao, and Guohui Li. 2024. Multi-scale context-aware user interest learning for behavior pattern modeling. InInternational Conference on Database Systems for Advanced Applications. Springer, 333–342

    2024

  8. [8]

    Ruining He and Julian McAuley. 2016. Ups and downs: Modeling the visual evolution of fashion trends with one-class collaborative filtering. Inproceedings of the 25th international conference on world wide web. 507–517

    2016

  9. [9]

    Xiangnan He, Kuan Deng, Xiang Wang, Yan Li, Yong-Dong Zhang, and Meng Wang. 2020. LightGCN: Simplifying and Powering Graph Convolution Net- work for Recommendation. InProceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval. 639–648

    2020

  10. [10]

    Haoji Hu and Xiangnan He. 2019. Sets2Sets: Learning from Sequential Sets with Neural Networks. InProceedings of the 25th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining, KDD. 1491–1499

    2019

  11. [11]

    Haoji Hu, Xiangnan He, Jinyang Gao, and Zhi-Li Zhang. 2020. Modeling Person- alized Item Frequency Information for Next-basket Recommendation. InProceed- ings of the 43rd International ACM SIGIR conference on research and development in Information Retrieval, SIGIR. 1071–1080

    2020

  12. [12]

    Kingma and Jimmy Ba

    Diederik P. Kingma and Jimmy Ba. 2015. Adam: A Method for Stochastic Opti- mization. In3rd International Conference on Learning Representations, ICLR

    2015

  13. [13]

    Hourun Li, Yifan Wang, Zhiping Xiao, Jia Yang, Changling Zhou, Ming Zhang, and Wei Ju. 2025. DisCo: graph-based disentangled contrastive learning for cold-start cross-domain recommendation. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 12049–12057

    2025

  14. [14]

    Kexin Li, Chengjiang Long, Shengyu Zhang, Xudong Tang, Zhichao Zhai, Kun Kuang, and Jun Xiao. 2024. CoreRec: A Counterfactual Correlation Inference for Next Set Recommendation. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 8661–8669

    2024

  15. [15]

    Ming Li, Yuanna Liu, Sami Jullien, Mozhdeh Ariannezhad, Andrew Yates, Mo- hammad Aliannejadi, and Maarten de Rijke. 2024. Are we really achieving better beyond-accuracy performance in next basket recommendation?. InProceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval. 924–934

    2024

  16. [16]

    Peiyang Liu, Xi Wang, Ziqiang Cui, and Wei Ye. 2025. Queries Are Not Alone: Clustering Text Embeddings for Video Search. InProceedings of the 48th Inter- national ACM SIGIR Conference on Research and Development in Information Retrieval. 874–883

    2025

  17. [17]

    Peiyang Liu, Jinyu Yang, Lin Wang, Sen Wang, Yunlai Hao, and Huihui Bai. 2023. Retrieval-Based Unsupervised Noisy Label Detection on Text Data. InProceed- ings of the 32nd ACM International Conference on Information and Knowledge Management. 4099–4104

    2023

  18. [18]

    Yuanna Liu, Ming Li, Mohammad Aliannejadi, and Maarten de Rijke. 2025. Repeat- Bias-Aware Optimization of Beyond-Accuracy Metrics for Next Basket Recom- mendation. InEuropean Conference on Information Retrieval. Springer, 214–229

    2025

  19. [19]

    Laurens van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-SNE. Journal of machine learning research9, Nov (2008), 2579–2605

    2008

  20. [20]

    Hongyi Mao, Mingsong Mao, and Fuhua Mao. 2024. Ranking on user–item heterogeneous graph for Ecommerce next basket recommendations.Knowledge- Based Systems296 (2024), 111863

    2024

  21. [21]

    Zhonghong Ou, Xiao Zhang, Yifan Zhu, Shuai Lyu, Jiahao Liu, and Tu Ao. 2025. LS-TGNN: Long and Short-Term Temporal Graph Neural Network for Session- Based Recommendation. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 39. 12426–12434

    2025

  22. [22]

    Seongmin Park, Mincheol Yoon, Minjin Choi, and Jongwuk Lee. 2025. Temporal Linear Item-Item Model for Sequential Recommendation. InProceedings of the Eighteenth ACM International Conference on Web Search and Data Mining. 354– 362

    2025

  23. [23]

    Bo Peng, Zhiyun Ren, Srinivasan Parthasarathy, and Xia Ning. 2023. M2: Mixed Models With Preferences, Popularities and Transitions for Next-Basket Recom- mendation.IEEE Trans. Knowl. Data Eng.35, 4 (2023), 4033–4046

    2023

  24. [24]

    Tianhao Peng, Haitao Yuan, Yongqi Zhang, Yuchen Li, Peihong Dai, Qunbo Wang, Senzhang Wang, and Wenjun Wu. 2025. Tagrec: Temporal-aware graph con- trastive learning with theoretical augmentation for sequential recommendation. IEEE Transactions on Knowledge and Data Engineering(2025)

    2025

  25. [25]

    Yuqi Qin, Pengfei Wang, and Chenliang Li. 2021. The World is Binary: Contrastive Learning for Denoising Next Basket Recommendation. InSIGIR ’21: The 44th International ACM SIGIR Conference on Research and Development in Information Retrieval. 859–868

    2021

  26. [26]

    Mostafa Rahmani, James Caverlee, and Fei Wang. 2023. Incorporating time in sequential recommendation models. InProceedings of the 17th ACM Conference on Recommender Systems. 784–790

    2023

  27. [27]

    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, WWW. 811–820

    2010

  28. [28]

    Yanyan Shen, Baoyuan Ou, and Ranzhen Li. 2022. MBN: Towards Multi-Behavior Sequence Modeling for Next Basket Recommendation.ACM Trans. Knowl. Discov. Data16, 5 (2022), 81:1–81:23

    2022

  29. [29]

    Yunxiao Shi, Haoning Shang, Xing Zi, Wujiang Xu, Yue Feng, and Min Xu

  30. [30]

    InProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing

    Answering narrative-driven recommendation queries via a retrieve–rank paradigm and the OCG-agent. InProceedings of the 2025 Conference on Empirical Methods in Natural Language Processing. 13192–13213

    2025

  31. [31]

    Yunxiao Shi, Wujiang Xu, Zhang Zeqi, Xing Zi, Qiang Wu, and Min Xu. 2025. PersonaX: A recommendation agent-oriented user modeling framework for long behavior sequence. InFindings of the Association for Computational Linguistics: ACL 2025. 5764–5787

    2025

  32. [32]

    Yehjin Shin, Jeongwhan Choi, Hyowon Wi, and Noseong Park. 2024. An atten- tive inductive bias for sequential recommendation beyond the self-attention. In Proceedings of the AAAI conference on artificial intelligence, Vol. 38. 8984–8992

    2024

  33. [33]

    Ting-Ting Su, Chang-Dong Wang, Wu-Dong Xi, Jian-Huang Lai, and Philip S Yu

  34. [34]

    Hierarchical Alignment With Polar Contrastive Learning for Next-Basket Recommendation.IEEE Transactions on Knowledge & Data Engineering36, 01 (2024), 199–210

    2024

  35. [35]

    Jiaxi Tang and Ke Wang. 2018. Personalized Top-N Sequential Recommendation via Convolutional Sequence Embedding.WSDM ’18: Proceedings of the Eleventh ACM International Conference on Web Search and Data Mining, 573

    2018

  36. [36]

    Shengxian Wan, Yanyan Lan, Pengfei Wang, Jiafeng Guo, Jun Xu, and Xueqi Cheng. 2015. Next Basket Recommendation with Neural Networks. InPoster Proceedings of the 9th ACM Conference on Recommender Systems, RecSys, Vol. 1441

    2015

  37. [37]

    Pengfei Wang, Jiafeng Guo, Yanyan Lan, Jun Xu, Shengxian Wan, and Xueqi Cheng. 2015. Learning Hierarchical Representation Model for NextBasket Rec- ommendation. InProceedings of the 38th International ACM SIGIR Conference on Research and Development in Information Retrieval. 403–412

    2015

  38. [38]

    Sheng, Mehmet Orgun, and Long- bing Cao

    Shoujin Wang, Liang Hu, Yan Wang, Quan Z. Sheng, Mehmet Orgun, and Long- bing Cao. 2020. Intention2Basket: A Neural Intention-Driven Approach for Dynamic Next-Basket Planning. InProceedings of the Twenty-Ninth International Joint Conference on Artificial Intelligence. 2333–2339

    2020

  39. [39]

    Sheng, Mehmet A

    Shoujin Wang, Liang Hu, Yan Wang, Quan Z. Sheng, Mehmet A. Orgun, and Longbing Cao. 2020. Intention Nets: Psychology-Inspired User Choice Behavior Modeling for Next-Basket Prediction. InThe Thirty-Fourth AAAI Conference on Artificial Intelligence. 6259–6266

    2020

  40. [40]

    Xin Wang, Hong Chen, Yuwei Zhou, Jianxin Ma, and Wenwu Zhu. 2023. Dis- entangled Representation Learning for Recommendation.IEEE Transactions on Pattern Analysis and Machine Intelligence (TPAMI)45, 1 (2023), 408–424

    2023

  41. [41]

    Yifan Wang, Yiping Song, Shuai Li, Chaoran Cheng, Wei Ju, Ming Zhang, and Sheng Wang. 2022. Disencite: Graph-based disentangled representation learning for context-specific citation generation. InProceedings of the AAAI conference on artificial intelligence, Vol. 36. 11449–11458

    2022

  42. [42]

    Yuling Wang, Xiao Wang, Xiangzhou Huang, Yanhua Yu, Haoyang Li, Mengdi Zhang, Zirui Guo, and Wei Wu. [n. d.]. Intent-aware Recommendation via Disen- tangled Graph Contrastive Learning. InProceedings of the Thirty-Second Interna- tional Joint Conference on Artificial Intelligence. 2343–2351

  43. [43]

    Chuyuan Wei, Baojie Yuan, Chuanhao Hu, Jinzhe Li, Chang-Dong Wang, and Mohsen Guizani. 2024. Knowledge-Aware Intent-Guided Contrastive Learning for Next-basket Recommendation.IEEE Transactions on Emerging Topics in Computational Intelligence(2024)

    2024

  44. [44]

    Lianghao Xia, Chao Huang, Yong Xu, and Jian Pei. 2022. Multi-behavior sequen- tial recommendation with temporal graph transformer.IEEE Transactions on Knowledge and Data Engineering35, 6 (2022), 6099–6112

    2022

  45. [45]

    Shuo Xiao, Jingtao Zhang, Chaogang Tang, and Zhenzhen Huang. 2025. Frequency-Domain Disentanglement-Fusion and Dual Contrastive Learning for Sequential Recommendation. InProceedings of the 34th ACM International SIGIR ’26, July 20–24, 2026, Melbourne, VIC, Australia Zhiying Deng et al. Conference on Information and Knowledge Management, CIKM. 3498–3508

    2025

  46. [46]

    Jiawei Yao and Juhua Hu. 2024. Dual-disentangled deep multiple clustering. In Proceedings of the 2024 SIAM International Conference on Data Mining (SDM). SIAM, 679–687

    2024

  47. [47]

    Haochao Ying, Fuzhen Zhuang, Fuzheng Zhang, Yanchi Liu, Guandong Xu, Xing Xie, Hui Xiong, and Jian Wu. 2018. Sequential Recommender System based on Hi- erarchical Attention Networks. InProceedings of the Twenty-Seventh International Joint Conference on Artificial Intelligence, IJCAI. 3926–3932

    2018

  48. [48]

    Feng Yu, Qiang Liu, Shu Wu, Liang Wang, and Tieniu Tan. 2016. A Dynamic Recurrent Model for Next Basket Recommendation. InProceedings of the 39th International ACM SIGIR conference on Research and Development in Information Retrieval, SIGIR. 729–732

    2016

  49. [49]

    Yalin Yu, Enneng Yang, Guibing Guo, Linying Jiang, and Xingwei Wang. 2023. Basket Representation Learning by Hypergraph Convolution on Repeated Items for Next-basket Recommendation.. InIJCAI. 2415–2422

    2023

  50. [50]

    Shengzhe Zhang, Liyi Chen, Chao Wang, Shuangli Li, and Hui Xiong. 2024. Tem- poral graph contrastive learning for sequential recommendation. InProceedings of the AAAI conference on artificial intelligence, Vol. 38. 9359–9367

    2024

  51. [51]

    Sen Zhao, Wei Wei, Ding Zou, and Xianling Mao. 2022. Multi-view intent disen- tangle graph networks for bundle recommendation. InProceedings of the AAAI conference on artificial intelligence, Vol. 36. 4379–4387

    2022

  52. [52]

    Yangtao Zhou, Hua Chu, Qingshan Li, Jianan Li, Shuai Zhang, Feifei Zhu, Jingzhao Hu, Luqiao Wang, and Wanqiang Yang. 2025. Dual-tower model with semantic perception and timespan-coupled hypergraph for next-basket recommendation. Neural Networks184 (2025), 107001

    2025