Pith. sign in

REVIEW 4 major objections 6 minor 56 references

AdaptGOT: A Pre-trained Model for Adaptive Contextual POI Representation Learning

T0 review · 4 major / 6 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Adaptive POI embeddings beat fixed pretrained baselines by up to 23.8%

desk verdict Empirical POI embedding work that deserves a review, but the headline 'task-adaptive without retraining' is not backed by the equations, and the JS-divergence loss is underspecified. read the letter →

arxiv 2506.21612 v1 pith:RDWMRKLK submitted 2025-06-21 cs.CL cs.AIcs.IR

Xiaobin Ren , Xinyu Zhu , Kaiqi Zhao This is my paper · ORCID
classification cs.CLcs.AIcs.IR
keywords point-of-interestembeddingpretrainedrepresentationlearningmixtureofexpertsgraphattentionsubgraphsamplingnextPOIrecommendationJensen-Shannondivergencegeo-contextual
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

AdaptGOT is a pre-trained model for point-of-interest (POI) embeddings that claims a single embedding can serve multiple downstream tasks by adapting itself per task. The model fuses three contexts — geographical location, co-occurrence patterns, and user-generated text — and combines four subgraph sampling strategies so that neighborhoods reflect proximity, density, importance, and category. A mixture-of-experts aggregator then weights these contextual subgraphs differently for each task without retraining, while a Jensen-Shannon divergence term is meant to keep subgraph representations topologically faithful to the full POI graph. On Foursquare (New York, Tokyo) and Yelp (Louisiana, Nevada) data, the paper reports consistent Recall@K improvements over fixed pretrained embeddings across next-POI recommendation, next-category prediction, and POI recommendation, with the largest stated gain being a 23.8% Rec@5 improvement over SpaBERT under FPMC in New York. A sympathetic reader would take the central thesis to be that task-adaptive, multi-context pretraining beats fixed one-size-fits-all POI embeddings.

What carries the argument

The load-bearing mechanism is the adaptive encoder–decoder built on a sparse mixture-of-experts (MoE) layer. Given four candidate subgraphs produced by KNN, density-based, importance, and category-aware sampling, the MoE gating function computes a sparse, task-dependent weighting, and the model minimizes the Jensen–Shannon divergence between the weighted subgraph feature distribution and the original graph feature distribution to preserve topological consistency. Around this sit the GOT attention module, which folds relative geographical position and normalized co-occurrence into the attention score while BERT-encoded review text forms node features, and a masked self-supervised objective that reconstructs text, geo, and co-occurrence signals. The expressive-power argument (Theorems 1–3) ties the design to the 1-WL isomorphism test: multi-context features reduce label conflicts and increase entropy, which is why mixed subgraphs are claimed to beat single-context message passing.

What would settle it

Run AdaptGOT with the JS divergence loss replaced by a fixed constant, keeping everything else identical, and compare the learned MoE gate weights and downstream Recall@K. If the gate weights and recall are unchanged, the JS term is not carrying the topological-consistency mechanism; conversely, if the term measurably changes which experts are selected, the claimed mechanism has a verifiable effect. A second check is to define $P(G_s)$ and $P(G)$ explicitly on a small POI graph and recompute Eq. (23) to see whether the minimized value actually tracks topological differences.

Watch

Extended reading notes

Core claim

On its own terms, AdaptGOT claims to establish that pre-trained POI representations are not required to be fixed. By jointly encoding geographical, co-occurrence, and textual signals in a modified graph attention mechanism and letting a sparse mixture-of-experts layer pick different subgraph topologies for different tasks, the model produces embeddings that outperform fixed pretrained baselines (CTLE, SpaBERT) and random or fixed initializations across three POI tasks on two real-world datasets. The paper also claims theoretical support: multiple contexts and multiple subgraphs increase the discriminative power and entropy of node embeddings under the 1-WL isomorphism test, which is offered as the reason mixed sampling improves expressiveness. Ablations attribute the gains to each component — removing the adaptive aggregator, the attention module, or any single context degrades recall — and a zero-shot cross-city experiment reports an average recall increase of 19.7%.

Load-bearing premise

The whole 'adaptive and topologically consistent' claim rests on the assumption that the Jensen–Shannon divergence in Eq. (23) is a well-defined objective between identifiable distributions of subgraph and original-graph features, yet the paper never defines how $P(G_s)$ and $P(G)$ are computed; if that term is not well-defined, the adaptive module reduces to a learned gating network and the stated mechanism is not doing the claimed work.

Editorial extensions

If this is right

  • One pre-trained POI embedding can be adapted at inference time to tasks with different contextual emphases, so downstream models no longer need task-specific fine-tuning of the embedding layer.
  • Cross-city transfer improves: in the paper's zero-shot experiment, AdaptGOT raises average recall by 19.7% over the fixed baselines, suggesting embeddings trained on one city can serve another.
  • Each context channel earns its keep: ablations show removing geography, co-occurrence, text, or the adaptive aggregator lowers recall, so multi-context integration is doing measurable work.
  • The expressive-power argument implies that adding more POI contexts or more subgraphs should keep improving label discrimination up to a point, with diminishing returns after four subgraphs as observed.

Reading between the lines

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

  • If the JS-divergence term turns out not to be well-defined, the practical contribution likely survives as a task-adaptive gating network over multi-context subgraphs; the loss could be replaced by a simpler diversity or entropy regularizer.
  • The same mixed-sampling-plus-MoE recipe could transfer to other geo-entity problems such as land-use classification, region-function identification, or route prediction, where no single neighborhood definition is correct for all queries.
  • A direct test of the 1-WL argument would be to measure embedding entropy or oversmoothing (for example, average cosine similarity between neighboring node embeddings) across subgraph counts; the theory predicts less oversmoothing with more contexts.
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.

Referee Report

4 major / 6 minor

Summary. The paper proposes AdaptGOT, a pre-training framework for POI embeddings. It combines mixed subgraph sampling (KNN, density, importance, and category-aware strategies), a GOT representation module that fuses geographical, co-occurrence, and textual information, a modified GOT attention mechanism, and an MoE-based 'adaptive representation aggregator' claimed to adjust POI representations across downstream tasks without retraining by minimizing a Jensen-Shannon divergence between subgraph and original graph distributions. The model is trained self-supervised with text, geography, and co-occurrence reconstruction losses. Experiments on Foursquare and Yelp evaluate next-POI recommendation, next-category prediction, and POI recommendation, with ablations, sensitivity analyses, and a zero-shot cross-city transfer test; the paper reports consistent recall improvements over SpaBERT, CTLE, and several other baselines.

Significance. If its mechanism claims were fully supported, AdaptGOT would be a useful contribution to POI representation learning: the multi-context sampling and GOT attention are sensible design choices, and the evaluation breadth (multiple datasets, tasks, ablations, cross-city transfer) is a strength. However, the central 'adaptive across tasks without retraining' claim is not implemented in the forward pass as written, and the JS-divergence objective is not operationalized. The empirical comparison also lacks variance reporting and selects hyperparameters on the evaluation metrics. The contribution is therefore currently under-supported, though these issues do not by themselves disprove the reported recall gains.

major comments (4)
  1. [§4.4, Eqs. (20)–(22)] The forward pass in Eqs. (20)–(22) computes h'_i from the GOT-attention output z_i and learned gating weights only; no task identifier, task loss, or downstream query appears in these equations. This contradicts the claim in Section 1 and Section 4.4 that the MoE aggregator 'dynamically adjusts POI representations across tasks without retraining': a fixed pretrained POI receives the same h'_i for next-POI recommendation, category prediction, and POI recommendation unless the downstream model fine-tunes the embedding. Please state precisely what 'adaptive across tasks' means operationally, or revise the claim to describe a fixed multimodal embedding.
  2. [§4.4, Eq. (23)] The JS-divergence term that is supposed to preserve 'topological consistency' is not operationalized. P(G_s) and P(G) are never defined or estimated, no equation links D_JS to h'_i or to a graph-topological quantity, and the claim that minimizing Eq. (23) aligns subgraph representations with global structure is therefore unsupported. As written, the adaptive representation aggregator reduces to a learned per-node gating network, and the JS loss contributes nothing identifiable to the mechanism. Define the distributions, give an estimator, and show how the loss affects the gating or expert outputs.
  3. [§3.2, Theorems 1–3] The theoretical claims supporting the 'expressive power' contribution are not established in the manuscript. The proofs are deferred to an anonymous external README, and the statements in Eqs. (4)–(6) are not formal: P_conflict and entropy are not defined on the same probability space, and the proof sketch ('feature space size grows exponentially') does not by itself imply the claimed 1-WL discriminative-power result. Either provide complete proofs in the paper or reframe these statements as intuition and empirical motivation.
  4. [§5.1.4 and Table 2] The empirical evidence for superiority is weakened by the absence of error bars and significance tests, and by the fact that key hyperparameters (number of subgraphs, heads, neighbors) are selected on the same Recall@K metrics used for the final comparison. Table 2 also contains cases where the advertised gain is zero or negligible, for example LA-Next POI Rec@5 under LSTPM: GOT 0.078 versus SpaBERT 0.078. Please report means over repeated runs with variance, use a validation-based selection protocol, and identify where the improvement is statistically reliable.
minor comments (6)
  1. [Throughout] The term 'GOT' is used ambiguously for both the model (AdaptGOT) and the representation module (GOT representation); Table 2's 'GOT' row should be labeled consistently to avoid confusion.
  2. [Abstract and template] There are numerous typos and formatting artifacts: 'POl' in the Abstract, 'onGeographical location' in the Abstract, the repeated 'the' in Section 4.4, and the placeholder ACM template with 2018 copyright and 'Conference acronym ’XX'.
  3. [§5.2.1] The text refers to 'Los Angeles' when the dataset column is labeled 'LA' for Louisiana; please make the city/state labels consistent.
  4. [Eqs. (8) and (28)] Equation (28) uses the notation g ∈ G(h_i) without defining the distribution over gates, and Eq. (8) uses N_den both as a set and as its cardinality; please clean up the notation.
  5. [Footnote 1] The footnote pointing to an anonymous README for the proofs and NDCG results is not an adequate substitute for material that should be in the paper or a stable repository; include the proofs and full metric tables in the manuscript.
  6. [Table 5] The zero-shot cross-city table does not state which source/target city is used for each column; without this information, the transfer claim is hard to interpret.

Circularity Check

0 steps flagged · score 4.0 of 10

The empirical headline results are partly circular because the model configuration is selected on the same Rec@K metrics used to substantiate superiority; the separate 'adaptive across tasks' claim is unsupported but not a circular derivation.

full rationale

The central empirical comparison is against external baselines (CTLE, SpaBERT, GETNext, FPMC, LSTPM, etc.) on two real-world datasets, so the claim of superior performance is not circular by construction. The only concrete circularity I can exhibit is the model-selection step: the number of subgraphs, number of heads, and other hyperparameters are tuned on Recall@K/NDCG for the same downstream tasks and datasets later used as evidence, without a described independent holdout; this makes the headline numbers partly fitted rather than predicted. This is a real circularity in the validation chain, but it does not reduce the whole architecture to its inputs because the model still encodes genuinely external data. I also flag, as non-circular correctness concerns, that Section 4.4's claim of task-adaptive representations 'without retraining' is unsupported: Eqs. 20-22 contain no task identifier or downstream query, so the same POI receives the same embedding across tasks; and Eq. 23's JS divergence is not operationalized because P(G_s) and P(G) are never defined or estimated. Theorems 1-3 defer proofs to an anonymous link, and reference [41] is by a co-author but is used only as a baseline, not as load-bearing justification. These issues reduce confidence but do not constitute a circular derivation of the empirical results.

Assumptions & free parameters 5 free parameters · 4 assumptions · 0 invented entities

The model rests on a graph formulation, on informal expressiveness theorems, and on a set of sampling hyperparameters. Most free parameters are either unreported or selected from test-set sensitivity studies, which is the main ledger item dragging down reproducibility and soundness.

free parameters (5)
  • Number of subgraphs/expert paths = 4
    Section 5.5.1: 'Performance peaked with four subgraphs... we chose four subgraphs as optimal.' This is a test-set choice on the same Recall metric used for evaluation, so the reported numbers are partly selected.
  • Number of attention heads = 2
    Section 5.5.3 and Figure 6: 2 heads selected as superior on the evaluation task; used in final model per Section 5.1.4.
  • Number of neighbor connections k in pre-training = 5 (with 15 recommended in sensitivity)
    Section 5.1.4 sets k=5, but Section 5.5.2 reports 15 neighbors outperform in the majority of cases on Yelp LA. The final choice is inconsistent and appears tuned on the evaluation metric.
  • KDE bandwidth b in density-based sampling = not reported
    Equation (8) defines density with a Gaussian kernel and bandwidth b, but b is never specified, making the density neighborhood non-reproducible.
  • Smoothing constant gamma in importance and category sampling = not reported
    Equations (9) and (10) add a 'small value' gamma to prevent zero denominators; the value is not given.
assumptions (4)
  • domain assumption POI graph formulation: POIs are nodes of a graph whose edges are learnable and determined by node features (Section 3.2, following [50]).
    The whole framework is built on this graph representation, but the paper does not validate that learnable edges plus sampled neighborhoods capture the relevant POI relations.
  • ad hoc to paper Multi-context enrichment strictly increases 1-WL discriminative power and embedding entropy (Theorems 1-3).
    The theorems are asserted with informal proofs (Section 3.2.3-3.2.4) and the detailed proofs are deferred to an anonymous repository; the model design leans on this claim.
  • domain assumption Mixed user review text aggregated per POI is a meaningful textual representation (Section 4.2.3).
    Reviews are concatenated across users and encoded by BERT; no analysis shows this concatenation preserves the signal used for importance sampling.
  • domain assumption Co-occurrence representation without time encoding supports zero-shot cross-city transfer (Section 4.2.1).
    The paper asserts this rationale for choosing co-occurrence, but the cross-city experiment (Table 5) is small and has no spatial or temporal normalization check.

how reviews work

0 comments
Cite this review

Pith. "Pith review of AdaptGOT: A Pre-trained Model for Adaptive Contextual POI Representation Learning." pith.science (2026). https://pith.science/paper/RDWMRKLK

@misc{pith2026250621612,
  author       = {Pith},
  title        = {Pith review of: AdaptGOT: A Pre-trained Model for Adaptive Contextual POI Representation Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/RDWMRKLK}},
  note         = {Machine review of arXiv:2506.21612}
}
read the original abstract

Currently, considerable strides have been achieved in Point-of-Interest (POI) embedding methodologies, driven by the emergence of novel POI tasks like recommendation and classification. Despite the success of task-specific, end-to-end models in POI embedding, several challenges remain. These include the need for more effective multi-context sampling strategies, insufficient exploration of multiple POI contexts, limited versatility, and inadequate generalization. To address these issues, we propose the AdaptGOT model, which integrates both the (Adapt)ive representation learning technique and the Geographical-Co-Occurrence-Text (GOT) representation with a particular emphasis on Geographical location, Co-Occurrence and Textual information. The AdaptGOT model comprises three key components: (1) contextual neighborhood generation, which integrates advanced mixed sampling techniques such as KNN, density-based, importance-based, and category-aware strategies to capture complex contextual neighborhoods; (2) an advanced GOT representation enhanced by an attention mechanism, designed to derive high-quality, customized representations and efficiently capture complex interrelations between POIs; and (3) the MoE-based adaptive encoder-decoder architecture, which ensures topological consistency and enriches contextual representation by minimizing Jensen-Shannon divergence across varying contexts. Experiments on two real-world datasets and multiple POI tasks substantiate the superior performance of the proposed AdaptGOT model.

Figures

Figures reproduced from arXiv: 2506.21612 by the authors.

Figure 1
Figure 1. (a) Visualization of POIs on Yelp Nevada dataset. (b) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Architecture (a), (b), (c) and expressive power (d) of [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Ablation study on mixed sampling and GOT [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗
Figures from the paper (2 more)
Figure 5
Figure 5. Figure 5: Sensitivity studies on different number of [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Sensitivity study on different heads numbers. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

56 extracted references · 28 canonical work pages

  1. [1]

    Chris Burges, Tal Shaked, Erin Renshaw, Ari Lazier, Matt Deeds, Nicole Hamilton, and Greg Hullender. 2005. Learning to rank using gradient descent. InProceedings of the 22nd international conference on Machine learning . 89–96

    2005

  2. [2]

    Buru Chang, Yonggyu Park, Donghyeon Park, Seongsoon Kim, and Jaewoo Kang. 2018. Content-Aware Hierarchical Point-of-Interest Embedding Model for Successive POI Recommendation. In Proceedings of the 27th International Joint Conference on Artificial Intelligence (Stockholm, Sweden) (IJCAI’18). AAAI Press, 3301–3307

    work page 2018

  3. [3]

    Jie Chen, Tengfei Ma, and Cao Xiao. 2018. FastGCN: Fast Learning with Graph Convolutional Networks via Importance Sampling. (1 2018). http://arxiv.org/ abs/1801.10247

    arXiv 2018

  4. [4]

    Jianfei Chen, Jun Zhu, and Le Song. 2017. Stochastic training of graph con- volutional networks with variance reduction. arXiv preprint arXiv:1710.10568 (2017)

    arXiv 2017

  5. [5]

    Meng Chen, Yan Zhao, Yang Liu, Xiaohui Yu, and Kai Zheng. 2020. Modeling spatial trajectories with attribute representation learning. IEEE transactions on knowledge and data engineering 34, 4 (2020), 1902–1914

    work page 2020

  6. [6]

    Meng Chen, Lei Zhu, Ronghui Xu, Yang Liu, Xiaohui Yu, and Yilong Yin. 2021. Embedding Hierarchical Structures for Venue Category Representation. ACM Trans. Inf. Syst. 40, 3, Article 57 (nov 2021), 29 pages. doi:10.1145/3478285

    work page doi:10.1145/3478285 2021

  7. [7]

    Wei-Lin Chiang, Xuanqing Liu, Si Si, Yang Li, Samy Bengio, and Cho-Jui Hsieh

  8. [8]

    Weilin Cong, Rana Forsati, Mahmut Kandemir, and Mehrdad Mahdavi. 2020. Minimal variance sampling with provable guarantees for fast training of graph neural networks. In Proceedings of the 26th ACM SIGKDD International Conference on Knowledge Discovery & Data Mining . 1393–1403

    2020

Show all 56 references
  1. [9]

    Yue Cui, Hao Sun, Yan Zhao, Hongzhi Yin, and Kai Zheng. 2021. Sequential- Knowledge-Aware Next POI Recommendation: A Meta-Learning Approach.ACM Trans. Inf. Syst. 40, 2, Article 23 (sep 2021), 22 pages. doi:10.1145/3460198

    2021 doi

  2. [10]

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. 2018. BERT: Pre-training of Deep Bidirectional Transformers for Language Understanding. CoRR abs/1810.04805 (2018). arXiv:1810.04805 http://arxiv.org/abs/1810.04805

    2018 arXiv

  3. [11]

    Ruixue Ding, Boli Chen, Pengjun Xie, Fei Huang, Xin Li, Qiang Zhang, and Yao Xu. 2023. A Multi-Modal Geographic Pre-Training Method. arXiv preprint arXiv:2301.04283 (2023)

    2023 arXiv

  4. [12]

    Shanshan Feng, Gao Cong, Bo An, and Yeow Meng Chee. 2017. Poi2vec: Geo- graphical latent representation for predicting future visitors. In Proceedings of the AAAI Conference on Artificial Intelligence , Vol. 31

    2017

  5. [13]

    Hazelton

    Martin L. Hazelton. 2014. Kernel Smoothing. Wiley StatsRef: Statistics Reference Online

    2014

  6. [14]

    Xiangnan He, Lizi Liao, Hanwang Zhang, Liqiang Nie, Xia Hu, and Tat-Seng Chua. 2017. Neural collaborative filtering. In Proceedings of the 26th international conference on world wide web . 173–182

    2017

  7. [15]

    Wenbing Huang, Tong Zhang, Yu Rong, and Junzhou Huang. 2018. Adaptive sam- pling towards fast graph representation learning. Advances in neural information processing systems 31 (2018)

    2018

  8. [16]

    Yannis Karmim, Elias Ramzi, Raphaël Fournier-S’niehotta, and Nicolas Thome

  9. [17]

    Minseok Kim, Jinoh Oh, Jaeyoung Do, and Sungjin Lee. 2022. Debiasing Neighbor Aggregation for Graph Neural Network in Recommender Systems. International Conference on Information and Knowledge Management, Proceedings , 4128–4132

    2022

  10. [18]

    Diederik P Kingma and Jimmy Ba. 2014. Adam: A method for stochastic opti- mization. arXiv preprint arXiv:1412.6980 (2014)

    2014 arXiv

  11. [19]

    Dejiang Kong and Fei Wu. 2018. HST-LSTM: A hierarchical spatial-temporal long-short term memory network for location prediction.. In IJCAI, Vol. 18. 2341–2347

    2018

  12. [20]

    Yehuda Koren, Robert Bell, and Chris Volinsky. 2009. Matrix factorization tech- niques for recommender systems. Computer 42, 8 (2009), 30–37

    2009

  13. [21]

    Yang Li, Yadan Luo, Zheng Zhang, Shazia Sadiq, and Peng Cui. 2019. Context- aware attention-based data augmentation for POI recommendation. In 2019 IEEE 35th International Conference on Data Engineering Workshops (ICDEW) . IEEE, 177–184

    2019

  14. [22]

    Zekun Li, Jina Kim, Yao-Yi Chiang, and Muhao Chen. 2022. SpaBERT: A Pre- trained Language Model from Geographic Data for Geo-Entity Representation. arXiv:2210.12213 [cs.CL]

    2022 arXiv

  15. [23]

    Defu Lian, Yongji Wu, Yong Ge, Xing Xie, and Enhong Chen. 2020. Geography- aware sequential location recommendation. In Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining . 2009– 2019

    2020

  16. [24]

    Yan Lin, Huaiyu Wan, Shengnan Guo, and Youfang Lin. 2021. Pre-training context and time aware location embeddings from spatial-temporal trajectories for user next location prediction. In Proceedings of the AAAI Conference on Artificial Intelligence, Vol. 35. 4241–4248

    2021

  17. [25]

    Fan Liu, Zhiyong Cheng, Lei Zhu, Zan Gao, and Liqiang Nie. 2021. Interest-aware Message-Passing GCN for Recommendation. In WWW ’21: The Web Conference 2021, Virtual Event / Ljubljana, Slovenia, April 19-23, 2021 , Jure Leskovec, Marko Grobelnik, Marc Najork, Jie Tang, and Leil...

    2021

  18. [26]

    Xin Liu, Yong Liu, and Xiaoli Li. 2016. Exploring the context of locations for personalized location recommendations.. In IJCAI. 1188–1194

    2016

  19. [27]

    Steven Loria et al. 2018. textblob Documentation. Release 0.15 2, 8 (2018), 269

    2018

  20. [28]

    Saeed Masoudnia and Reza Ebrahimpour. 2014. Mixture of experts: a literature survey. Artificial Intelligence Review 42 (2014), 275–293

    2014

  21. [29]

    Christopher Morris, Gaurav Rattan, and Petra Mutzel. 2020. Weisfeiler and Leman go sparse: Towards scalable higher-order graph embeddings. In Advances in Neu- ral Information Processing Systems 33: Annual Conference on Neural Information Processing Systems 2020, NeurIPS 2020, ...

    2020

  22. [30]

    Christopher Morris, Martin Ritzert, Matthias Fey, William L Hamilton, Jan Eric Lenssen, Gaurav Rattan, and Martin Grohe. 2019. Weisfeiler and leman go neural: Higher-order graph neural networks. In Proceedings of the AAAI conference on artificial intelligence, Vol. 33. 4602–4609

    2019

  23. [31]

    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

  24. [32]

    N Shazeer, A Mirhoseini, K Maziarz, A Davis, Q Le, G Hinton, and J Dean. 2017. The sparsely-gated mixture-of-experts layer. Outrageously large neural networks (2017)

    2017

  25. [33]

    Ke Sun, Tieyun Qian, Tong Chen, Yile Liang, Nguyen Hung, and Hongzhi Yin

  26. [34]

    Petar Veličković, Guillem Cucurull, Arantxa Casanova, Adriana Romero, Pietro Lio, and Yoshua Bengio. 2017. Graph attention networks. arXiv preprint arXiv:1710.10903 (2017)

    2017 arXiv

  27. [35]

    Haotao Wang, Ziyu Jiang, Yuning You, Yan Han, Gaowen Liu, Jayanth Srinivasa, Ramana Rao Kompella, and Zhangyang Wang. 2023. Graph Mixture of Experts: Learning on Large-Scale Graphs with Explicit Diversity Modeling. (4 2023). http://arxiv.org/abs/2304.02806

    2023 arXiv

  28. [36]

    Xinfeng Wang, Fumiyo Fukumoto, Jin Cui, Yoshimi Suzuki, Jiyi Li, and Dongjin Yu

  29. [37]

    Xiang Wang, Xiangnan He, Meng Wang, Fuli Feng, and Tat-Seng Chua. 2019. Neural graph collaborative filtering. In Proceedings of the 42nd international ACM SIGIR conference on Research and development in Information Retrieval . 165–174

    2019

  30. [38]

    Keyulu Xu, Weihua Hu, Jure Leskovec, and Stefanie Jegelka. 2019. How Powerful are Graph Neural Networks?. In 7th International Conference on Learning Rep- resentations, ICLR 2019, New Orleans, LA, USA, May 6-9, 2019 . OpenReview.net. https://openreview.net/forum?id=ryGs6iA5Km

    2019

  31. [39]

    Jingsong Yang, Guanzhou Han, Deqing Yang, Jingping Liu, Yanghua Xiao, Xiang Xu, Baohua Wu, and Shenghua Ni. 2023. M3pt: A multi-modal model for poi tag- ging. In Proceedings of the 29th ACM SIGKDD Conference on Knowledge Discovery and Data Mining. 5382–5392

    2023

  32. [40]

    Kang Yang, Ruiyun Yu, Bingyang Guo, and Jie Li. 2024. Interaction Subgraph Sequential Topology-Aware Network for Transferable Recommendation. IEEE Trans. Knowl. Data Eng. 36, 10 (2024), 5221–5233. doi:10.1109/TKDE.2024.3384965

    2024

  33. [41]

    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 (<conf-loc>, <city>Madrid</city>, <cou...

    2022

  34. [42]

    Di Yao, Chao Zhang, Jianhui Huang, and Jingping Bi. 2017. Serm: A recurrent model for next location prediction in semantic trajectories. In Proceedings of the 10 AdaptGOT: A Pre-trained Model for Adaptive Contextual POI Representation Learning Conference acronym ’XX, June 03–0...

    2017

  35. [43]

    Kai-Lang Yao and Wu-Jun Li. 2021. Blocking-based Neighbor Sampling for Large-scale Graph Neural Networks. https://cs.nju.edu.cn/lwj/

    2021

  36. [44]

    Feiyu Yin, Yong Liu, Zhiqi Shen, Lisi Chen, Shuo Shang, and Peng Han. 2023. Next POI Recommendation with Dynamic Graph and Explicit Dependency. In Thirty-Seventh AAAI Conference on Artificial Intelligence, AAAI 2023, Thirty- Fifth Conference on Innovative Applications of Artif...

    2023

  37. [45]

    Hanqing Zeng, Hongkuan Zhou, Ajitesh Srivastava, Rajgopal Kannan, and Viktor Prasanna. 2019. Graphsaint: Graph sampling based inductive learning method. arXiv preprint arXiv:1907.04931 (2019)

    2019 arXiv

  38. [46]

    Hao Zhang, Siyi Wei, Xiaojiao Hu, Ying Li, and Jiajie Xu. 2020. On accurate POI recommendation via transfer learning. Distributed and Parallel Databases 38 (2020), 585–599

    2020

  39. [47]

    Qianru Zhang, Lianghao Xia, Xuheng Cai, Siu-Ming Yiu, Chao Huang, and Christian S Jensen. 2024. Graph augmentation for recommendation. In 2024 IEEE 40th International Conference on Data Engineering (ICDE) . IEEE, 557–569

    2024

  40. [48]

    Zhenning Zhang, Boxin Du, and Hanghang Tong. 2022. SuGeR: A Subgraph- based Graph Convolutional Network Method for Bundle Recommendation. In Proceedings of the 31st ACM International Conference on Information & Knowledge Management, Atlanta, GA, USA, October 17-21, 2022 , Moha...

    2022

  41. [49]

    Lingxiao Zhao, Wei Jin, Leman Akoglu, and Neil Shah. 2021. From stars to subgraphs: Uplifting any GNN with local structure awareness. arXiv preprint arXiv:2110.03753 (2021)

    2021 arXiv

  42. [50]

    Zhehan Zhao, Lu Bai, Lixin Cui, Ming Li, Yue Wang, Lixiang Xu, and Edwin R Hancock. 2024. ENADPool: The Edge-Node Attention-based Differentiable Pool- ing for Graph Neural Networks. arXiv preprint arXiv:2405.10218 (2024)

    2024 arXiv

  43. [51]

    Difan Zou, Ziniu Hu, Yewen Wang, Song Jiang, Yizhou Sun, and Quanquan Gu

  44. [56]

    (11 2019)

    Ladies:Layer-Dependent Importance Sampling for Training Deep and Large Graph Convolutional Networks. (11 2019). Received 20 February 2007; revised 12 March 2009; accepted 5 June 2009 11

    2019

  45. [2019]

    In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining

    Cluster-gcn: An efficient algorithm for training deep and large graph convolutional networks. In Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery & data mining . 257–266

  46. [2020]

    Proceedings of the AAAI Conference on Artificial Intelligence 34 (04 2020), 214–221

    Where to Go Next: Modeling Long- and Short-Term User Preferences for Point-of-Interest Recommendation. Proceedings of the AAAI Conference on Artificial Intelligence 34 (04 2020), 214–221. doi:10.1609/aaai.v34i01.5353

    2020 doi

  47. [2023]

    In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval

    EEDN: Enhanced Encoder-Decoder Network with Local and Global Context Learning for POI Recommendation. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval . 383–392

  48. [2024]

    ITEM: Improving Training and Evaluation of Message-Passing based GNNs for top-k recommendation. Trans. Mach. Learn. Res. 2024 (2024). https: //openreview.net/forum?id=9B6LM2uoEs

    2024

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.