Pith. sign in

REVIEW 3 major objections 5 minor 2 cited by

Harnessing Large Language Models for Group POI Recommendations

T0 review · 3 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read LLMGPR uses a fine-tuned large language model to recommend next places for groups, beating specialist group recommenders by 4–9%.

desk verdict A genuinely new LLM-based group POI recommender with consistent gains, but a possible temporal leakage in the evaluation protocol needs to be resolved before those gains are credible. read the letter →

arxiv 2411.13415 v2 pith:253WBRDR submitted 2024-11-20 cs.IR

Jing Long , Liang Qu , Junliang Yu , Tong Chen , Quoc Viet Hung Nguyen , Hongzhi Yin This is my paper · ORCID
classification cs.IR
keywords groupPOIrecommendationlargelanguagemodelsQLoRAself-supervisedlearningnextpredictionpreferenceaggregationlocation-basedsocialnetworkssemantic-enhancedtokens
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 large language models (LLMs), properly adapted, can outperform dedicated group point-of-interest (POI) recommenders. The proposed framework, LLMGPR, turns POIs into semantic tokens, fine-tunes an LLM with a QLoRA adapter to encode check-in sequences, and adds a second adapter that aggregates individual members' representations into a group representation. A self-supervised task predicting the purpose of a check-in sequence (business trip, tourism, etc.) supplies extra supervision to counter the extreme sparsity of group check-in data. If the claims hold, LLMGPR improves next-POI ranking accuracy over the best existing group recommender by average margins of 7.09% on Foursquare, 4.39% on Weeplace, and 9.01% on Gowalla, and also beats individual-level LLM recommenders on sequence representation quality.

What carries the argument

The central machinery is the semantic-enhanced POI token, a special token whose embedding is initialized by prompting a pre-trained LLM with a POI's name, category, description, reviews, coordinates, and address, then averaged from the final-layer token embeddings. These POI tokens are combined with word tokens and spatio-temporal difference signals, processed by Llama3-8b with a QLoRA sequencing adapter to produce check-in sequence representations. A second QLoRA aggregation adapter, fed with the set of member representations, produces an aggregated member representation that is added to the group sequence representation with a weight alpha. A purpose-prediction self-supervised task pre-trains the sequencing adapter and POI embeddings using 11 LLM-generated purpose labels for short sequences, giving the model semantic supervision beyond the sparse group check-ins.

What would settle it

Construct or obtain a test set of independently verified group outings (e.g., groups that explicitly visited together, or user-declared sessions), run LLMGPR and MICL on that test set, and check whether the 4–9% HR@k/NDCG@k advantage persists; if the gap vanishes or reverses, the benchmark result is an artifact of the co-location group definition.

Watch

Extended reading notes

Core claim

The central claim is that group POI recommendation can be reframed as a sequence-understanding and preference-aggregation problem solvable by a large language model, provided the model is given three components: POI tokens whose embeddings encode semantic text about each place, a QLoRA sequencing adapter that aligns the frozen LLM with check-in sequence representation, and a separate aggregation adapter that combines member-level sequence representations into an enhanced group representation. The paper further claims that a self-supervised purpose-prediction task, trained on LLM-generated labels for short check-in sequences, substantially enriches these representations and mitigates group-level data sparsity. All evidence is empirical: held-out next-POI ranking accuracy on three public LBSN datasets, where LLMGPR outperforms the strongest group baseline MICL and the strongest individual LLM baseline LLM4POI across HR@k and NDCG@k.

Load-bearing premise

The load-bearing premise is that a group is correctly defined as socially connected users who happen to visit the same venue at the same time; if co-location is often coincidence rather than a shared plan, the model is trained on a proxy for group behavior and the benchmark gains may not reflect real group outings.

Editorial extensions

If this is right

  • If LLMGPR's gains are real, group POI recommenders no longer need to rely only on sparse group check-in frequencies; semantic and contextual signals from individual check-ins can be transferred through the LLM and its adapters.
  • The purpose-prediction SSL task provides a reusable way to inject latent trip intent into sequence representations, which could improve next-POI accuracy in cold-start and low-data regimes beyond the group setting.
  • The architecture's separation of a frozen LLM, a sequencing adapter, and an aggregation adapter means the framework can be swapped to any base LLM with sufficient resources, not only Llama3-8b.
  • The reported cold-start experiments suggest the method degrades less than baselines when groups have fewer than ten check-ins, implying practical value for real deployment where new groups are common.
  • The approach is the first LLM-based group POI recommender, so it sets a new baseline for future work combining LLMs with group preference aggregation.

Reading between the lines

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

  • Because the group labels are constructed by co-location of socially connected users, the benchmark may reward models that predict co-presence patterns rather than genuine group decision-making; an independent validation with explicitly declared group outings would be needed to confirm transfer to real settings.
  • The semantic POI token initialization from LLM text embeddings is a general technique that could be applied to other ID-based recommenders to relieve cold-start, independent of the group setting.
  • The purpose-prediction supervision relies on LLM-generated labels, whose 92% agreement with manual labels was checked on only 20% of sequences; extending that manual check to all sequences or to a held-out city would test whether the SSL signal is robust.
  • The aggregation adapter's treatment of members as an unordered set of representations could be extended to model member roles or influence asymmetries, which the current averaging-plus-residual design does not explicitly capture.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper proposes LLMGPR, a group POI recommendation framework built on Llama3-8b. It introduces semantic-enhanced POI tokens initialized by prompting the LLM, a QLoRA sequencing adapter that maps check-in sequences to vector representations, an aggregation adapter that combines individual member representations into a group representation, and a self-supervised purpose-prediction task with 11 LLM-generated purpose labels. The framework is evaluated on Foursquare, Weeplace, and Gowalla under a leave-one-out next-POI protocol, reporting improvements over six group baselines and seven individual baselines, plus ablations and cold-start experiments.

Significance. If the reported results survive a corrected evaluation protocol, this would be a worthwhile contribution: it is one of the first LLM-based frameworks for group POI recommendation, the adapter architecture is sensible and clearly separates sequence encoding from preference aggregation, and the authors include ablations that isolate the contribution of each component. I also credit the authors for evaluating against multiple strong group baselines and for checking cold-start performance. The main uncertainties are the unresolved temporal masking of member histories in the group evaluation, the absence of variance or significance reporting, and the co-presence proxy used to define groups; each of these can be addressed in revision without changing the overall architecture.

major comments (3)
  1. [Section 4.1; Algorithm 1 (lines 19-26)] The evaluation protocol does not specify whether member check-in sequences X_u are truncated at the group's held-out timestamp. Because group activities are constructed from members' individual check-ins (Section 4.1), the member sequences used in lines 20-22 can contain the very POI that serves as the group's ground-truth next POI. The aggregation adapter is then trained and evaluated with the answer present in its input, which alone could explain the 4.4-9.0% improvements over MICL in Table 2. Please state explicitly how member histories are masked at both training and test time; if they are not masked, the experiments must be rerun with a temporal split that removes the group test POI, and any later check-ins, from each member's input sequence before the comparison.
  2. [Section 4.3; Tables 2-5] All reported numbers are single point estimates with no standard deviations, confidence intervals, or significance tests. The margins over the best baseline are small in several cells (e.g., HR@5 0.3182 vs 0.3068 on Foursquare), so the abstract's claim of 'significantly' enhanced accuracy is not supported by the evidence as presented. Please report means and standard deviations over multiple independent runs and, if appropriate, paired significance tests for the headline comparisons.
  3. [Section 4.1] The group ground truth in all datasets is defined as sets of socially connected users who visit the same venue at the same time, i.e., co-presence, not validated group decisions. Since this proxy is used for every dataset and every experiment, the reported improvements may characterize co-presence prediction rather than real group outing recommendation. Please add an explicit discussion of this construct-validity limitation and, ideally, a sanity analysis (e.g., filtering to repeated co-visits or a small human-labeled sample of intentional group events) to show the findings transfer to genuine group decisions.
minor comments (5)
  1. [Section 4.2] The baseline descriptions for GroupIM and CubeRec are identical; the GroupIM entry should describe its mutual-information maximization framework rather than the hypercube-based representation attributed to CubeRec.
  2. [Section 3.4; Algorithm 1] The training sets D, D_ssl, and D_group are not all formally defined before first use; please add explicit definitions in the notation section or at the start of Section 3.4.
  3. [Section 4.5; Table 4] The variant name LLMGPR-FT is confusing because the caption and text describe it as 'using the LLM directly without fine-tuning'; consider renaming the variants to avoid the apparent contradiction between 'FT' and 'without fine-tuning'.
  4. [Figure 3] The subplot labels and axis captions are missing or too small to read; please provide a clearer figure with labeled axes and a legend.
  5. [References] Reference [35] attributes Llama to Touvron et al., but the paper uses Llama3-8b; please verify that the citation corresponds to the correct model release.

Circularity Check

1 steps flagged · score 6.0 of 10

Group co-presence construction places the target POI inside member-sequence inputs, potentially making LLMGPR's accuracy gains a copying artifact rather than learned group preference.

  1. self definitional [Section 4.1 (Datasets and Evaluation Protocols) + Algorithm 1, lines 19-26]
    "we assume if a set of users who are connected on the social network visit the same venue at the same time, then they are regarded as members of a group, and the corresponding activities are group activities. ... for each of the check-in sequences, the last check-in POI is for testing ... for X_uk ∈ X(K_n) do e_uk = Θ_s(X_uk, E_poi); end for"

    The held-out group target is the last check-in of the group sequence. By the dataset-construction rule quoted above, that target is also a check-in of every member, so it is contained in X_uk unless member histories are truncated at the group's test timestamp. Algorithm 1 feeds the full X_uk to Θ_s and Equation 9 adds the aggregated member representation to the group embedding scored in Equation 5. Nothing in the evaluation section states such truncation, so the ground-truth POI token is available as an input token; the model can rank it by copying or attending to a token already in its input. The 4.4-9.0% gains over baselines therefore reduce, in part, to an input-output overlap created by the group definition.

full rationale

Most of LLMGPR's machinery is not circular: the next-POI objective is optimized against an external held-out ranking target, the semantic POI embeddings are initialized from Llama3-8b and then fine-tuned, the GPT-4 purpose labels are auxiliary and were manually spot-checked at 92%, and the QLoRA adapters are ordinary fine-tuning components. The decisive circularity risk is in the benchmark construction: groups are defined as users who co-visit the same venue at the same time, so group ground-truth check-ins are also member check-ins. Algorithm 1's aggregation branch consumes member sequences, and the paper never specifies truncating those sequences at the group test timestamp. If no such masking is applied, the target POI is literally an input token, making the reported improvements partially forced by construction. This is a conditional but load-bearing reduction; if the authors can confirm temporal masking of member histories, the central result would again rest on an external held-out target. The group-construction procedure is cited to the authors' own CubeRec paper, but that citation is not the load-bearing step; the load-bearing step is the co-presence definition combined with unmasked member inputs.

Assumptions & free parameters 6 free parameters · 5 assumptions · 2 invented entities

The framework's central claim depends on pretrained LLM knowledge, a heuristic definition of groups, and LLM-generated purpose labels. The directly fitted knobs are the aggregation weight alpha, QLoRA rank r, and evaluation or preprocessing choices such as the 500-candidate set and the 10-interaction filter.

free parameters (6)
  • alpha (aggregation weight) = 0.7
    Blends the group sequence embedding with the aggregated member embedding in Eq. 9; tuned on validation, with sensitivity reported in Sec. 4.7.
  • QLoRA rank r = 16
    Rank of the low-rank sequencing and aggregation adapters; tuned in Sec. 4.7; accuracy plateaus for r > 16.
  • Purpose label taxonomy (L=11) = 11 categories
    The set of trip-purpose classes for the SSL task in Sec. 3.3 is author-defined and determines the purpose prediction head.
  • Candidate set size = 500
    Evaluation ranks each ground-truth POI among the 500 nearest unvisited POIs (Sec. 4.1); this shapes every reported metric.
  • Short-sequence split threshold = 5 days
    Sequences are split when consecutive check-ins are more than five days apart (Sec. 3.3), determining which sequences receive purpose labels.
  • Minimum interaction filter = 10
    Users and POIs with fewer than 10 interactions are removed (Sec. 4.1), changing dataset statistics and sparsity.
assumptions (5)
  • domain assumption Llama3-8b provides useful semantic and world knowledge for POI text.
    Invoked in Sec. 2.3 and 3.1.1 to initialize POI embeddings and to classify trip purposes; if the pretrained model lacks this knowledge, the main mechanism weakens.
  • domain assumption Co-located social connections at the same venue and time constitute a group.
    Sec. 4.1 uses this heuristic to construct all group check-in data; the task definition depends on it.
  • domain assumption LLM-generated purpose labels are accurate ground truth for the SSL task.
    Sec. 3.3 generates labels for all short sequences with GPT-4 and manually checks 20%, reporting 92% accuracy; the remaining 80% are unverified.
  • ad hoc to paper Averaging final-layer token embeddings yields meaningful POI and sequence representations.
    Used for POI initialization and Eq. 4; no theoretical justification, only empirical validation through ablations.
  • domain assumption The next POI is among the 500 nearest unvisited POIs in the same region.
    Sec. 4.1 restricts candidates to 500 nearest POIs, citing location sensitivity; all metrics depend on this candidate set.
invented entities (2)
  • Semantic-enhanced POI token
    purpose: Lets the LLM treat each POI as a special token initialized from textual attributes and fine-tuned through the recommendation loss (Sec. 3.1.1).
    The token exists only inside the model; no external falsifiable handle such as a predicted measurement is provided, only downstream accuracy.
  • Purpose embedding matrix and 11 trip-purpose labels
    purpose: Provides the SSL target for enriching check-in sequence representations (Sec. 3.3).
    The taxonomy is author-designed; a 20% manual check gives limited external evidence, but the label set itself is not independently validated.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Harnessing Large Language Models for Group POI Recommendations." pith.science (2026). https://pith.science/paper/253WBRDR

@misc{pith2026241113415,
  author       = {Pith},
  title        = {Pith review of: Harnessing Large Language Models for Group POI Recommendations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/253WBRDR}},
  note         = {Machine review of arXiv:2411.13415}
}
read the original abstract

The rapid proliferation of Location-Based Social Networks (LBSNs) has underscored the importance of Point-of-Interest (POI) recommendation systems in enhancing user experiences. While individual POI recommendation methods leverage users' check-in histories to provide personalized suggestions, they struggle to address scenarios requiring group decision-making. Group POI recommendation systems aim to satisfy the collective preferences of multiple users, but existing approaches face two major challenges: diverse group preferences and extreme data sparsity in group check-in data. To overcome these challenges, we propose LLMGPR, a novel framework that leverages large language models (LLMs) for group POI recommendations. LLMGPR introduces semantic-enhanced POI tokens and incorporates rich contextual information to model the diverse and complex dynamics of group decision-making. To further enhance its capabilities, we developed a sequencing adapter using Quantized Low-Rank Adaptation (QLoRA), which aligns LLMs with group POI recommendation tasks. To address the issue of sparse group check-in data, LLMGPR employs an aggregation adapter that integrates individual representations into meaningful group representations. Additionally, a self-supervised learning (SSL) task is designed to predict the purposes of check-in sequences (e.g., business trips and family vacations), thereby enriching group representations with deeper semantic insights. Extensive experiments demonstrate the effectiveness of LLMGPR, showcasing its ability to significantly enhance the accuracy and robustness of group POI recommendations.

Figures

Figures reproduced from arXiv: 2411.13415 by the authors.

Figure 1
Figure 1. (a) Example of individual POI recommendation. (b) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. The overview of our proposed LLMGPR. these embeddings will be further fine-tuned during the following group POI recommendation task with specific objectives. 3.1.2 Generating Candidate Group POI. Given all POI embeddings, we design another prompt shown as follows to generate the representation of the check-in sequence. Generate an embedding for the provided check-in sequence, which has multiple check-in activities, … view at source ↗
Figure 3
Figure 3. Hyperparameter sensitivity. Impact of 𝛼. We experiment on 𝛼 ∈ {0, 0.1, 0.3, 0.5, 0.7, 0.9, 1}. The lowest accuracy is obtained if the aggregated members’ preference is not applied (𝛼 = 0), showing the significance of the strategy for enhancing group representations. At the start, the recommendation accuracy increases with the increase of 𝛼. However, the accuracy will decline if knowledge of members’ individual prefe… 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

  1. Think2Go: Generative Next POI Recommendation with LLM Reasoning

    cs.IR 2026-07 conditional novelty 4.0 of 10

    Think2Go couples SFT and RL-based reasoning in one LLM, with KDE- and reward-gap-based advantage calibration, and reports state-of-the-art Acc@1 on NYC, Tokyo, and California check-in data.

  2. 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

62 extracted references · 39 canonical work pages · cited by 2 Pith papers

  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]

    Da Cao, Xiangnan He, Lianhai Miao, Yahui An, Chao Yang, and Richang Hong

  3. [3]

    Tong Chen, Hongzhi Yin, Jing Long, Quoc Viet Hung Nguyen, Yang Wang, and Meng Wang. 2022. Thinking inside the box: learning hypercube representations for group recommendation. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval . 1664–1673

    2022

  4. [4]

    Z. Chen, H. Cao, H. Wang, F. Xu, and Y. Li. 2020. Will You Come Back / Check-in Again? Understanding Characteristics Leading to Urban Revisitation and Re- check-in. Proceedings of the ACM on Interactive Mobile Wearable and Ubiquitous Technologies (2020)

    work page 2020

  5. [5]

    Cheng, H

    C. Cheng, H. Yang, M. R. Lyu, and I. King. 2013. Where You Like to Go Next: Successive Point-of-Interest Recommendation.. In International Joint Conference on Artificial Intelligence

    work page 2013

  6. [6]

    Tim Dettmers, Artidoro Pagnoni, Ari Holtzman, and Luke Zettlemoyer. 2024. Qlora: Efficient finetuning of quantized llms. Advances in Neural Information Processing Systems 36 (2024)

    2024

  7. [7]

    Shanshan Feng, Haoming Lyu, Caishun Chen, and Yew-Soon Ong. 2024. Where to Move Next: Zero-shot Generalization of LLMs for Next POI Recommendation. arXiv preprint arXiv:2404.01855 (2024)

    arXiv 2024

  8. [8]

    Xinyi Gao, Tong Chen, Yilong Zang, Wentao Zhang, Quoc Viet Hung Nguyen, Kai Zheng, and Hongzhi Yin. 2023. Graph condensation for inductive node representation learning. arXiv preprint arXiv:2307.15967 (2023)

    work page Pith review arXiv 2023

Show all 62 references
  1. [9]

    Xinyi Gao, Wentao Zhang, Tong Chen, Junliang Yu, Hung Quoc Viet Nguyen, and Hongzhi Yin. 2023. Semantic-aware node synthesis for imbalanced heterogeneous information networks. In Proceedings of the 32nd ACM International Conference on Information and Knowledge Management . 545–555

    2023

  2. [10]

    Lei Guo, Hongzhi Yin, Tong Chen, Xiangliang Zhang, and Kai Zheng. 2021. Hierarchical hyperedge embedding-based representation learning for group recommendation. ACM Transactions on Information Systems (TOIS) 40, 1 (2021), 1–27

    2021

  3. [11]

    Yupeng Hou, Junjie Zhang, Zihan Lin, Hongyu Lu, Ruobing Xie, Julian McAuley, and Wayne Xin Zhao. 2024. Large language models are zero-shot rankers for recommender systems. In European Conference on Information Retrieval. Springer, 364–381

    2024

  4. [12]

    Sirui Huang, Qian Li, Xiangmeng Wang, Dianer Yu, Guandong Xu, and Qing Li. 2024. Counterfactual Debasing for Multi-behavior Recommendations. In International Conference on Database Systems for Advanced Applications . Springer, 164–179

    2024

  5. [13]

    Sirui Huang, Qian Li, Haoran Yang, Dianer Yu, Qing Li, and Guandong Xu. [n. d.]. Causal Time-aware News Recommendations with Large Language Models. ACM Transactions on Information Systems ([n. d.])

  6. [14]

    Ting-Ji Huang, Jia-Qi Yang, Chunxu Shen, Kai-Qi Liu, De-Chuan Zhan, and Han-Jia Ye. 2024. Improving LLMs for Recommendation with Out-Of-Vocabulary Tokens. arXiv preprint arXiv:2406.08477 (2024)

    2024 arXiv

  7. [15]

    F. Jie, L. Yong, Z. Chao, F. Sun, and D. Jin. 2018. DeepMove: Predicting Human Mobility with Attentional Recurrent Networks. In the 2018 World Wide Web Conference

    2018

  8. [16]

    Krichene and S

    W. Krichene and S. Rendle. 2020. On Sampled Metrics for Item Recommendation. In KDD ’20: The 26th ACM SIGKDD Conference on Knowledge Discovery and Data Mining

    2020

  9. [17]

    Guohui Li, Qi Chen, Bolong Zheng, Hongzhi Yin, Quoc Viet Hung Nguyen, and Xiaofang Zhou. 2020. Group-based recurrent neural networks for POI recommendation. ACM Transactions on Data Science 1, 1 (2020), 1–18

    2020

  10. [18]

    Jiuqiang Li and Shilei Zhu. 2024. Multi-View Interactive Compromise Learning for Group Recommendation. In ICASSP 2024-2024 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP) . IEEE, 9396–9400

    2024

  11. [19]

    Peibo Li, Maarten de Rijke, Hao Xue, Shuang Ao, Yang Song, and Flora D Salim

  12. [20]

    Ranzhen Li, Yanyan Shen, and Yanmin Zhu. 2018. Next Point-of-Interest Rec- ommendation with Temporal and Multi-level Context Attention. 2018 IEEE International Conference on Data Mining (ICDM) (2018), 1110–1115

    2018

  13. [21]

    Yang Li, Tong Chen, Yadan Luo, Hongzhi Yin, and Zi Huang. 2021. Discovering collaborative signals for next POI recommendation with iterative Seq2Graph augmentation. In IJCAI. 1491–1497

    2021

  14. [22]

    D. Lian, C. Zhao, X. Xie, G. Sun, E. Chen, and Y. Rui. 2014. GeoMF: joint geo- graphical modeling and matrix factorization for point-of-interest recommendation . ACM

    2014

  15. [23]

    Xinyu Lin, Wenjie Wang, Yongqi Li, Shuo Yang, Fuli Feng, Yinwei Wei, and Tat- Seng Chua. 2024. Data-efficient Fine-tuning for LLM-based Recommendation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval . 365–374

    2024

  16. [24]

    X. Liu, Y. Liu, K. Aberer, and C. Miao. 2013. Personalized point-of-interest recommendation by mining users’ preference transition. In Acm International Conference on Conference on Information & Knowledge Management

    2013

  17. [25]

    Zhizhong Liu, Lingqiang Meng, Quan Z Sheng, Dianhui Chu, Jian Yu, and Xiaoyu Song. 2024. POI recommendation for random groups based on cooperative graph neural networks. Information Processing & Management 61, 3 (2024), 103676

    2024

  18. [26]

    Jing Long, Tong Chen, Quoc Viet Hung Nguyen, Guandong Xu, Kai Zheng, and Hongzhi Yin. 2023. Model-Agnostic Decentralized Collaborative Learning for On-Device POI Recommendation. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Inform...

    2023

  19. [27]

    Jing Long, Tong Chen, Quoc Viet Hung Nguyen, and Hongzhi Yin. 2023. Decen- tralized collaborative learning framework for next POI recommendation. ACM Transactions on Information Systems 41, 3 (2023), 1–25

    2023

  20. [28]

    Jing Long, Tong Chen, Guanhua Ye, Kai Zheng, Nguyen Quoc Viet Hung, and Hongzhi Yin. 2024. Physical Trajectory Inference Attack and Defense in Decen- tralized POI Recommendation. arXiv preprint arXiv:2401.14583 (2024)

    2024 arXiv

  21. [29]

    Jing Long, Guanhua Ye, Tong Chen, Yang Wang, Meng Wang, and Hongzhi Yin. 2024. Diffusion-based cloud-edge-device collaborative learning for next POI recommendations. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . 2026–2036

    2024

  22. [30]

    Y. Luo, Q. Liu, and Z. Liu. 2021. STAN: Spatio-Temporal Attention Network for Next Location Recommendation

    2021

  23. [31]

    Yifang Qin, Hongjun Wu, Wei Ju, Xiao Luo, and Ming Zhang. 2023. A Diffusion Model for POI Recommendation. ACM Transactions on Information Systems 42, 2 (Nov. 2023), 1–27. https://doi.org/10.1145/3624475

    2023 doi

  24. [32]

    Liang Qu, Huaisheng Zhu, Ruiqi Zheng, Yuhui Shi, and Hongzhi Yin. 2021. Im- gagn: Imbalanced network embedding via generative adversarial graph networks. In Proceedings of the 27th ACM SIGKDD Conference on Knowledge Discovery & Data Mining. 1390–1398

    2021

  25. [33]

    Aravind Sankar, Yanhong Wu, Yuhang Wu, Wei Zhang, Hao Yang, and Hari Sundaram. 2020. Groupim: A mutual information maximization framework for neural group recommendation. InProceedings of the 43rd International ACM SIGIR conference on research and development in Information Re...

    2020

  26. [34]

    Zahra Bahari Sojahrood and Mohammad Taleai. 2021. A POI group recommen- dation method in location-based social networks based on user influence. Expert Systems with Applications 171 (2021), 114593

    2021

  27. [35]

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. 2023. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971 (2023)

    2023 arXiv

  28. [36]

    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 systems 30 (2017)

    2017

  29. [37]

    Lucas Vinh Tran, Tuan-Anh Nguyen Pham, Yi Tay, Yiding Liu, Gao Cong, and Xiaoli Li. 2019. Interact and decide: Medley of sub-attention networks for effective group recommendation. In Proceedings of the 42nd International ACM SIGIR conference on research and development in info...

    2019

  30. [38]

    Xinglei Wang, Meng Fang, Zichao Zeng, and Tao Cheng. 2023. Where would i go next? large language models as human mobility predictors. arXiv preprint arXiv:2308.15197 (2023)

    2023 arXiv

  31. [39]

    Zhaobo Wang, Yanmin Zhu, Haobing Liu, and Chunyang Wang. 2022. Learning Graph-based Disentangled Representations for Next POI Recommendation. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (, Madrid, Spain,) (...

    2022

  32. [40]

    Weimer, A

    M. Weimer, A. Karatzoglou, Q. V. Le, and A. J. Smola. 2007. CoFiRank - Maximum Margin Matrix Factorization for Collaborative Ranking. In Neural Information Processing Systems

    2007

  33. [41]

    Yiqing Wu, Ruobing Xie, Yongchun Zhu, Fuzhen Zhuang, Xu Zhang, Leyu Lin, and Qing He. 2024. Personalized prompt for sequential recommendation. IEEE Transactions on Knowledge and Data Engineering (2024)

    2024

  34. [42]

    Min Xie, Hongzhi Yin, Hao Wang, Fanjiang Xu, Weitong Chen, and Sen Wang

  35. [43]

    Song Yang, Jiamou Liu, and Kaiqi Zhao. 2022. GETNext: Trajectory Flow Map Enhanced Transformer for Next POI Recommendation. In Proceedings of the 45th International ACM SIGIR Conference on Research and Development in Information Retrieval (Madrid, Spain) (SIGIR ’22). Associati...

    2022

  36. [44]

    Hongzhi Yin and Bin Cui. 2016. Spatio-temporal recommendation in social media . Springer

    2016

  37. [45]

    Hongzhi Yin, Bin Cui, Zi Huang, Weiqing Wang, Xian Wu, and Xiaofang Zhou

  38. [46]

    Hongzhi Yin, Bin Cui, Xiaofang Zhou, Weiqing Wang, Zi Huang, and Shazia Sadiq

  39. [47]

    Hongzhi Yin, Liang Qu, Tong Chen, Wei Yuan, Ruiqi Zheng, Jing Long, Xin Xia, Yuhui Shi, and Chengqi Zhang. 2024. On-Device Recommender Systems: A Comprehensive Survey. arXiv preprint arXiv:2401.11441 (2024)

    2024 arXiv

  40. [48]

    Hongzhi Yin, Qinyong Wang, Kai Zheng, Zhixu Li, Jiali Yang, and Xiaofang Zhou. 2019. Social influence-based group representation learning for group recommendation. In 2019 IEEE 35th International Conference on Data Engineering (ICDE). IEEE, 566–577

    2019

  41. [49]

    Hongzhi Yin, Qinyong Wang, Kai Zheng, Zhixu Li, and Xiaofang Zhou. 2020. Overcoming data sparsity in group recommendation.IEEE Transactions on Knowl- edge and Data Engineering 34, 7 (2020), 3447–3460

    2020

  42. [50]

    Xiaohan Yu, Li Zhang, Xin Zhao, Yue Wang, and Zhongrui Ma. 2024. RA-Rec: An Efficient ID Representation Alignment Framework for LLM-based Recom- mendation. arXiv preprint arXiv:2402.04527 (2024)

    2024 arXiv

  43. [51]

    ACM Transactions on Information Systems (TOIS) 35, 2 (2016), 1–44

    Joint modeling of user check-in behaviors for real-time point-of-interest recommendation. ACM Transactions on Information Systems (TOIS) 35, 2 (2016), 1–44

    2016

  44. [52]

    Jianyang Zhai, Xiawu Zheng, Chang-Dong Wang, Hui Li, and Yonghong Tian

  45. [53]

    Junwei Zhang, Min Gao, Junliang Yu, Lei Guo, Jundong Li, and Hongzhi Yin. 2021. Double-scale self-supervised hypergraph learning for group recommendation. In Proceedings of the 30th ACM international conference on information & knowledge management. 2557–2567

    2021

  46. [54]

    Qianru Zhang, Peng Yang, Junliang Yu, Haixin Wang, Xingwei He, Siu-Ming Yiu, and Hongzhi Yin. 2025. A survey on point-of-interest recommendation: Models, architectures, and security. IEEE Transactions on Knowledge and Data Engineering (2025)

    2025

  47. [55]

    Shijie Zhang, Wei Yuan, and Hongzhi Yin. 2023. Comprehensive privacy analysis on federated recommender system against attribute inference attacks. IEEE Transactions on Knowledge and Data Engineering (2023)

    2023

  48. [56]

    Wei Yuan, Chaoqun Yang, Quoc Viet Hung Nguyen, Lizhen Cui, Tieke He, and Hongzhi Yin. 2023. Interaction-level membership inference attack against feder- ated recommender systems. arXiv preprint arXiv:2301.10964 (2023)

    2023 arXiv

  49. [62]

    Bolong Zheng, Kai Zheng, Xiaokui Xiao, Han Su, Hongzhi Yin, Xiaofang Zhou, and Guohui Li. 2016. Keyword-aware continuous knn query on road networks. In 2016 IEEE 32Nd international conference on data engineering (ICDE) . IEEE, 871–882

    2016

  50. [2015]

    In Proceedings of the 23rd ACM international conference on Harnessing Large Language Models for Group POI Recommendations CIKM ’25, November 10–14, 2025, Seoul, Republic of Korea

    Joint modeling of users’ interests and mobility patterns for point-of-interest recommendation. In Proceedings of the 23rd ACM international conference on Harnessing Large Language Models for Group POI Recommendations CIKM ’25, November 10–14, 2025, Seoul, Republic of Korea. Mu...

    2025

  51. [2016]

    In Proceedings of the 25th ACM international on conference on information and knowledge management

    Learning graph-based poi embedding for location-based recommendation. In Proceedings of the 25th ACM international on conference on information and knowledge management. 15–24

  52. [2018]

    In The 41st International ACM SIGIR conference on research & development in information retrieval

    Attentive group recommendation. In The 41st International ACM SIGIR conference on research & development in information retrieval . 645–654

  53. [2023]

    In Proceedings of the 31st ACM International Conference on Multimedia

    Knowledge prompt-tuning for sequential recommendation. In Proceedings of the 31st ACM International Conference on Multimedia . 6451–6461

  54. [2024]

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

    Large language models for next point-of-interest recommendation. In Proceedings of the 47th International ACM SIGIR Conference on Research and Development in Information Retrieval . 1463–1472

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.