Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

Geography-Aware Large Language Models for Next POI Recommendation

T0 review · 4 major / 5 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 This paper claims that adding quadkey-Fourier geographic encodings and POI alignment projections to a fine-tuned LLM improves next-POI top-1 accuracy by about 15-17% over strong prior baselines.

desk verdict GA-LLM is a serious, well-engineered empirical paper that combines known components into a strong next-POI recommender, but the evaluation has an unaddressed leakage risk from precomputed POI embeddings and thin statistics that prevent full confidence in the headline numbers. read the letter →

arxiv 2505.13526 v1 pith:J442ZOXI submitted 2025-05-18 cs.IR cs.AI

Zhao Liu , Wei Liu , Huajie Zhu , Jianxing Yu , Jian Yin , Wang-Chien Lee , Shun Wang This is my paper · ORCID
classification cs.IRcs.AI
keywords nextPOIrecommendationlargelanguagemodelsgeographiccoordinateinjectionquadkeyencodinglearnableFourierpositionaltransitionalignmentcold-startLLMfine-tuning
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

The paper argues that large language models struggle at next-POI recommendation because they cannot reason about precise GPS coordinates and have no built-in knowledge of which places people tend to visit next. To fix this, GA-LLM feeds each location through a compact spatial encoder and injects transition-aware POI embeddings into the model's token space. On three real-world check-in datasets the model reports top-1 accuracy of 0.3919 on NYC, 0.3482 on TKY, and 0.2566 on CA, beating the strongest prior baselines by 16.19%, 14.73%, and 16.69%. The central point is that structured geographic and transition signals, rather than text descriptions of coordinates, let an LLM predict destinations it never saw in its prompt.

What carries the argument

The central mechanism is the pairing of two injection modules. GCIM converts each GPS point into a quadkey string (a base-4 quadtree tile identifier whose prefix length encodes the zoom level), builds overlapping n-grams from that string, applies self-attention, and adds a learnable Fourier positional encoding so the LLM receives a compact multi-scale spatial representation instead of a long string of coordinate tokens. PAM maps low-dimensional POI embeddings, taken from a pretrained sequential recommender such as MTNet, through an MLP into the LLM's token embedding space, so transition knowledge is available as input tokens during fine-tuning. The two modules together are what allow the LLM to predict a target POI that never appears in its prompt.

What would settle it

Retrain GA-LLM with POI embeddings computed only from the training split (or audit the released code for the split boundary) and compare Acc@1 on the three datasets; a large drop would indicate leakage, while a small drop would confirm the reported gains.

Watch

Extended reading notes

Core claim

GA-LLM establishes that next-POI recommendation can be improved by treating geography and POI transitions as structured inputs to an LLM rather than as raw text. GCIM discretizes GPS coordinates into hierarchical quadkey strings, enriches them with overlapping n-grams and self-attention, and applies a learnable Fourier positional encoding to produce a compact spatial token; PAM projects precomputed POI embeddings from a sequential model into the LLM's semantic space via an MLP. With Llama-2-7b fine-tuned on formatted check-in queries, the model reports Acc@1 of 0.3919 on NYC, 0.3482 on TKY, and 0.2566 on CA, outperforming the strongest prior baseline by 16.19%, 14.73%, and 16.69%, and reducing the average distance of wrong predictions on CA from 61.38 km to 37.63 km. The paper also reports that removing either module lowers accuracy, and that the geographic encoder alone improves cross-city cold-start accuracy.

Load-bearing premise

The reported accuracy gains rest on POI embeddings precomputed by a separate sequential model, and the paper never states that those embeddings were trained only on the training split; if the embedding model saw test check-ins, transition patterns from ground-truth POIs would leak into the LLM and inflate Acc@1.

Editorial extensions

If this is right

  • Next-POI top-1 accuracy can be raised by roughly 15-17% over prior best models on standard benchmarks, suggesting that LLM-based recommenders need structured geographic encoding rather than coordinate text.
  • Compact quadkey-Fourier representations reduce the number of tokens per check-in, making fine-tuning faster and per-query inference faster than a text-only LLM baseline that uses 200 historical check-ins per query.
  • GCIM alone enables cross-city cold-start transfer: a model trained on one city predicts POIs in another city more accurately than a text-only LLM, even when the target POI is absent from the input.
  • PAM improves prediction accuracy specifically for cases where the target POI does not appear in the prompt, and it is designed to work with different sequential or graph-based POI embedding sources.
  • Wrong predictions land closer to the true location on sparse datasets, indicating that the spatial encoding reduces geographically implausible LLM outputs.

Reading between the lines

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

  • If the accuracy gains survive a clean train/test split of the sequential model's POI embeddings, the same injection recipe could transfer to other spatial LLM tasks such as destination prediction or trajectory generation, where discrete grid encodings may outperform raw coordinate text.
  • Because GCIM relies on the Mercator projection and quadkey tiling, its transferability to non-Mercator or globe-spanning settings is untested; a direct experiment would evaluate GA-LLM on data from extreme latitudes or a different map projection.
  • The cross-city results hint that spatial structure is learnable independently of locale, but the ceiling of GA-LLM may be set by the external sequential model that produces POI embeddings; replacing MTNet with a simple bigram transition matrix would reveal how much of the gain comes from transition knowledge versus the LLM's own reasoning.
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 / 5 minor

Summary. The paper proposes GA-LLM, a framework that augments a fine-tuned LLM (Llama-2-7b-longlora-32k) with two components for next-POI recommendation: the Geographic Coordinate Injection Module (GCIM), which encodes GPS coordinates via quadkey-based hierarchical discretization, n-gram/self-attention, and learnable Fourier features, and the POI Alignment Module (PAM), which projects POI embeddings from a sequential model (e.g., MTNet) into the LLM semantic space. Experiments on Foursquare-NYC, Foursquare-TKY, and Gowalla-CA report state-of-the-art Acc@1 (0.3919, 0.3482, 0.2566), with ablations showing that both GCIM and PAM contribute to performance and that the approach improves cold-start and cross-city settings relative to LLM4POI. The paper also reports efficiency gains from compact spatial encoding.

Significance. If the reported results hold, GA-LLM would be a meaningful advance in applying LLMs to spatially grounded recommendation: it addresses a real limitation of text-only LLMs (poor spatial reasoning and hallucinated distant POIs) with two well-motivated, plug-in components, and it demonstrates consistent gains over strong sequential and graph baselines across three datasets. The ablation study is internally consistent, and the authors make the code available through an anonymous link. The main value lies in showing that explicit geographic encoding and POI-transition alignment can be injected into an LLM without representing every coordinate as verbose text, which is a practical contribution to the LLM-for-recommendation literature. The significance is conditioned, however, on resolving a critical data-split question: if the precomputed POI embeddings used by PAM were trained on the test portion of the check-in data, the reported accuracy gains would be inflated and the central claim would not stand.

major comments (4)
  1. [§4.3, §5.1, §5.5] The paper never states whether the MTNet POI embeddings consumed by PAM were trained only on the chronological 80% training split or on the full dataset. Section 4.3 and Section 5.5 describe PAM as using embeddings from sequential models such as MTNet, and Section 5.5 says the experiments use MTNet, but Section 5.1 only describes the 80/10/10 split for the LLM fine-tuning data. If MTNet was fit on all check-ins before the split, transition statistics involving test POIs would leak into the LLM through the PAM input, inflating Table 2 and Table 3. Please state the split used to train MTNet (and any GCIM/PAM pretraining mentioned in Section 5.6), retrain the embeddings on the training split only, or provide a leakage check such as comparing against randomly permuted POI embeddings.
  2. [§5.4, Table 4] The cross-city cold-start protocol is not explained. Table 4's header is ambiguous (it reads 'Model Trained on NYC Acc@1 TKY Acc@1 CA Acc@1'), and the text does not say whether rows or columns index the training city, how POI IDs and categories are matched between cities with disjoint POI vocabularies (4,981 vs. 7,833 vs. 9,690), or what 'trained on TKY and tested on NYC' means for the GA-LLM w/o PAM variant. Without a precise protocol, the cross-city generalization claim in Section 5.4 is not verifiable from the table.
  3. [§5.1, §5.2] Table 2 reports only a single Acc@1 value per dataset with no runs, variance, or error bars, and the significance statement ('p-value ≤ 0.05') omits the test used and the number of repetitions. Additionally, SeCor, LLMmove, and LLaRA are discussed in Section 2.2 as relevant LLM-based recommenders, but none appears in the comparison. Please either include these baselines or justify their exclusion from the main table; the current evaluation does not fully support the claim of state-of-the-art performance among LLM-based methods.
  4. [§5.5, §5.6, Figure 7] The efficiency comparison in Section 5.6 is not controlled: LLM4POI is stated to use 200 check-ins per query while GA-LLM uses fewer, so the reported shorter fine-tuning and inference times could simply reflect shorter input sequences rather than the efficiency of the proposed modules. Also, the statement in Section 5.6 that 'excluding pre-training for GCIM and PAM' is excluded from timing appears to conflict with Section 5.5, where PAM is described as a projector trained during LLM fine-tuning. Please clarify what pretraining is performed for GCIM/PAM, on which data split, and report token counts or input lengths per method alongside timing.
minor comments (5)
  1. [§5.1] The text contains typos: 'Dose GA-LLM outperforms' should read 'Does GA-LLM outperform', and 'which also exploiting an LLM like GA-LLM dose' should read 'which also exploits an LLM like GA-LLM does'.
  2. [Table 2] The baseline list in the text includes MTNet [Huang et al., 2024], but Table 2 has a row labeled 'MTNext'; please unify the naming.
  3. [Figure 3] The legend in Figure 3 uses 'GAM-LLM text-geo' and 'GAM-LLM w/o PAM'; the acronym should be 'GA-LLM' consistently.
  4. [Figure 6] The caption and text are inconsistent: the caption mentions 'LLM4POI and GA-LLM w/o GCIM' while the surrounding text discusses PAM's effect on predictions where the target POI is absent. Please align the description with the actual comparison.
  5. [§5.4, Table 4] The row/column orientation of Table 4 should be made explicit (e.g., rows as training city, columns as test city) and the table should note whether the numbers are Acc@1 values for the same model or for different model variants.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: GA-LLM is a constructive pipeline evaluated against external baselines; no derivation step reduces to its inputs.

full rationale

GA-LLM's derivation chain is not circular. GCIM takes raw GPS coordinates and applies quadkey discretization (from external GeoSAN work), n-gram self-attention, and a learnable Fourier projection; these are independent techniques and do not encode the target next-POI label by construction. PAM takes POI embeddings from a separately trained sequential/graph model (e.g., MTNet) and learns an MLP projection into the LLM's semantic space; the projection is trained during supervised LLM fine-tuning on the next-POI objective, so the model must still learn to choose the correct next POI from trajectory text plus these projected features. The reported Acc@1 numbers are compared against standard baselines on a chronological 80/10/10 split, and no equation in the paper reduces a prediction to a fitted input or to a self-citation. The only concern—the paper does not explicitly state that the MTNet POI embeddings used by PAM were trained only on the 80% training split—is a potential data-leakage/correctness issue rather than circularity: nothing in the paper's equations makes the prediction equal to the precomputed embeddings, and the paper does not claim to derive the embeddings from the target labels. Therefore no circular step can be quoted.

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

The paper introduces no new physical or conceptual entities. The main unresolved burdens are the unreported hyperparameters (zoom level, quadkey depth, n-gram size, gamma, dimensions) and the untested assumption that precomputed POI embeddings are leakage-free. These are typical for an applied ML paper, but they matter for the truth of the reported numbers.

free parameters (5)
  • quadkey zoom level l
    Controls the resolution of geographic discretization; the paper does not report the value used in the experiments.
  • number of quadkey hierarchical levels L
    Length of the quadkey-derived representation S; not reported.
  • overlapping n-gram size
    The n-gram width used to generate sequences from the quadkey string is not stated.
  • Fourier kernel width gamma
    Controls the spatial kernel width in the initialization of Ws; no value or sensitivity analysis is given.
  • GCIM/PAM projection dimensions
    Output dimensionalities for the learnable projections into the LLM semantic space are not specified.
assumptions (4)
  • domain assumption Mercator projection and quadtree tile discretization preserve meaningful spatial proximity for POI recommendation.
    GCIM relies on this mapping to convert GPS coordinates into quadkeys; if the discretization loses important spatial structure, the geographic encoding would be ineffective.
  • domain assumption POI transition relations captured by graph or sequential models can be transferred into the LLM's semantic space via a simple MLP projection.
    PAM's design assumes that a linear-ish projection of low-dimensional POI embeddings is sufficient to align them with the LLM's token space; no analysis or experiment isolates whether this mapping is the key factor.
  • domain assumption Extreme position bias Acc@1, predicting a single exact POI ID from the full vocabulary, is a valid proxy for next-POI recommendation quality.
    The evaluation frames recommendation as open-set question answering; this may not reflect standard ranking-based POI recommendation, and the paper never reports Recall@k or NDCG.
  • domain assumption Precomputed POI embeddings are trained only on the training split and do not see test-set check-ins.
    The paper does not document the training split or procedure for the embedding model used in PAM; if the model was trained on the full dataset, test labels leak into the LLM.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Geography-Aware Large Language Models for Next POI Recommendation." pith.science (2026). https://pith.science/paper/J442ZOXI

@misc{pith2026250513526,
  author       = {Pith},
  title        = {Pith review of: Geography-Aware Large Language Models for Next POI Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J442ZOXI}},
  note         = {Machine review of arXiv:2505.13526}
}
read the original abstract

The next Point-of-Interest (POI) recommendation task aims to predict users' next destinations based on their historical movement data and plays a key role in location-based services and personalized applications. Accurate next POI recommendation depends on effectively modeling geographic information and POI transition relations, which are crucial for capturing spatial dependencies and user movement patterns. While Large Language Models (LLMs) exhibit strong capabilities in semantic understanding and contextual reasoning, applying them to spatial tasks like next POI recommendation remains challenging. First, the infrequent nature of specific GPS coordinates makes it difficult for LLMs to model precise spatial contexts. Second, the lack of knowledge about POI transitions limits their ability to capture potential POI-POI relationships. To address these issues, we propose GA-LLM (Geography-Aware Large Language Model), a novel framework that enhances LLMs with two specialized components. The Geographic Coordinate Injection Module (GCIM) transforms GPS coordinates into spatial representations using hierarchical and Fourier-based positional encoding, enabling the model to understand geographic features from multiple perspectives. The POI Alignment Module (PAM) incorporates POI transition relations into the LLM's semantic space, allowing it to infer global POI relationships and generalize to unseen POIs. Experiments on three real-world datasets demonstrate the state-of-the-art performance of GA-LLM.

Figures

Figures reproduced from arXiv: 2505.13526 by the authors.

Figure 1
Figure 1. Illustration of text-only LLM limitations in POI recom [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the GA-LLM framework. The left illustrates the workflow, where user trajectories are transformed into queries for [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Cumulative distribution function (CDF) of distances be [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figures from the paper (1 more)
Figure 6
Figure 6. Figure 6: Proportion of successful predictions where the target POI [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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.

Reference graph

Works this paper leans on

38 extracted references · 33 canonical work pages · cited by 1 Pith paper

  1. [1]

    wav2vec 2.0: A framework for self-supervised learning of speech repre- sentations

    [Baevski et al., 2020] Alexei Baevski, Yuhao Zhou, Abdel- rahman Mohamed, and Michael Auli. wav2vec 2.0: A framework for self-supervised learning of speech repre- sentations. In NeurIPS,

    work page 2020

  2. [4]

    Lyu, and Irwin King

    [Cheng et al., 2013] Chen Cheng, Haiqin Yang, Michael R. Lyu, and Irwin King. Where you like to go next: Succes- sive point-of-interest recommendation. In IJCAI, pages 2605–2611. IJCAI/AAAI,

    work page 2013

  3. [8]

    [He et al., 2016] Jing He, Xin Li, Lejian Liao, Dandan Song, and William K. Cheung. Inferring a personalized next point-of-interest recommendation model with latent be- havior patterns. In Proceedings of the Thirtieth AAAI Con- ference on Artificial Intelligence , pages 137–143. AAAI Press,

    work page 2016

  4. [11]

    Self-attentive sequential recommendation

    [Kang, 2018] Wang-Cheng Kang. Self-attentive sequential recommendation. In ICDM, pages 197–206. IEEE Com- puter Society,

    work page 2018

  5. [13]

    Learnable fourier features for multi- dimensional spatial positional encoding

    [Li et al., 2021] Yang Li, Si Si, Gang Li, Cho-Jui Hsieh, and Samy Bengio. Learnable fourier features for multi- dimensional spatial positional encoding. In NeurIPS, pages 15816–15829,

    work page 2021

  6. [15]

    Geography-aware sequential lo- cation recommendation

    [Lian et al., 2020] Defu Lian, Yongji Wu, Yong Ge, Xing Xie, and Enhong Chen. Geography-aware sequential lo- cation recommendation. In The 26th ACM SIGKDD Con- ference on Knowledge Discovery and Data Mining, pages 2009–2019. ACM,

    work page 2020

  7. [16]

    Llara: Large language-recommendation assistant

    [Liao et al., 2024] Jiayi Liao, Sihang Li, Zhengyi Yang, Jiancan Wu, Yancheng Yuan, Xiang Wang, and Xiangnan He. Llara: Large language-recommendation assistant. In SIGIR, pages 1785–1795. ACM,

    work page 2024

  8. [17]

    Rella: Retrieval-enhanced large language models for lifelong sequential behavior comprehension in recommendation

    [Lin et al., 2024] Jianghao Lin, Rong Shan, Chenxu Zhu, Kounianhua Du, Bo Chen, Shigang Quan, Ruiming Tang, Yong Yu, and Weinan Zhang. Rella: Retrieval-enhanced large language models for lifelong sequential behavior comprehension in recommendation. In Proceedings of the ACM on Web Conference 2024, pages 3497–3508. ACM,

    work page 2024

Show all 38 references
  1. [18]

    Large language models enhanced sequential recommenda- tion for long-tail user and item

    [Liu et al., 2024] Qidong Liu, Xian Wu, Xiangyu Zhao, Yejing Wang, Zijian Zhang, Feng Tian, and Yefeng Zheng. Large language models enhanced sequential recommenda- tion for long-tail user and item. CoRR, abs/2405.20646,

    2024 arXiv

  2. [19]

    STAN: spatio-temporal attention network for next location recommendation

    [Luo et al., 2021] Yingtao Luo, Qiang Liu, and Zhaocheng Liu. STAN: spatio-temporal attention network for next location recommendation. In WWW, pages 2177–2185. ACM / IW3C2,

    2021

  3. [20]

    Learning transferable visual models from natural language supervi- sion

    [Radford et al., 2021] Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agar- wal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervi-...

    2021

  4. [21]

    Graph-flashback net- work for next location recommendation

    [Rao et al., 2022] Xuan Rao, Lisi Chen, Yong Liu, Shuo Shang, Bin Yao, and Peng Han. Graph-flashback net- work for next location recommendation. In The 28th ACM SIGKDD Conference on Knowledge Discovery and Data Mining, pages 1463–1471. ACM,

    2022

  5. [22]

    Factorizing person- alized markov chains for next-basket recommendation

    [Rendle et al., 2010] Steffen Rendle, Christoph Freuden- thaler, and Lars Schmidt-Thieme. Factorizing person- alized markov chains for next-basket recommendation. In Proceedings of the 19th International Conference on World Wide Web, pages 811–820. ACM,

    2010

  6. [25]

    Graphgpt: Graph instruction tuning for large language models

    [Tang et al., 2024] Jiabin Tang, Yuhao Yang, Wei Wei, Lei Shi, Lixin Su, Suqi Cheng, Dawei Yin, and Chao Huang. Graphgpt: Graph instruction tuning for large language models. In SIGIR, pages 491–500. ACM,

    2024

  7. [26]

    Llama 2: Open foundation and fine-tuned chat models

    [Touvron et al., 2023] Hugo Touvron, Louis Martin, Kevin Stone, Peter Albert, Amjad Almahairi, and et. Llama 2: Open foundation and fine-tuned chat models. CoRR, abs/2307.09288,

    2023 arXiv

  8. [27]

    Secor: Align- ing semantic and collaborative representations by large language models for next-point-of-interest recommenda- tions

    [Wang et al., 2024] Shirui Wang, Bohan Xie, Ling Ding, Xi- aoying Gao, Jianting Chen, and Yang Xiang. Secor: Align- ing semantic and collaborative representations by large language models for next-point-of-interest recommenda- tions. In Proceedings of the 18th ACM Conference o...

    2024

  9. [28]

    Llmrec: Large language mod- els with graph augmentation for recommendation

    [Wei et al., 2024] Wei Wei, Xubin Ren, Jiabin Tang, Qiny- ong Wang, Lixin Su, Suqi Cheng, Junfeng Wang, Dawei Yin, and Chao Huang. Llmrec: Large language mod- els with graph augmentation for recommendation. In Proceedings of the 17th ACM International Conference on Web Search ...

    2024

  10. [29]

    Personalized long- and short-term prefer- ence learning for next POI recommendation

    [Wu et al., 2022] Yuxia Wu, Ke Li, Guoshuai Zhao, and Xueming Qian. Personalized long- and short-term prefer- ence learning for next POI recommendation. IEEE Trans. Knowl. Data Eng., 34(4):1944–1957,

    2022

  11. [30]

    [Wu et al., 2024] Junda Wu, Cheng-Chun Chang, Tong Yu, Zhankui He, Jianing Wang, Yupeng Hou, and Julian J. McAuley. Coral: Collaborative retrieval-augmented large language models improve long-tail recommendation. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Di...

    2024

  12. [31]

    Spatio-temporal hypergraph learning for next POI recommendation

    [Yan et al., 2023] Xiaodong Yan, Tengwei Song, Yifeng Jiao, Jianshan He, Jiaotuan Wang, Ruopeng Li, and Wei Chu. Spatio-temporal hypergraph learning for next POI recommendation. In SIGIR, pages 403–412. ACM,

    2023

  13. [32]

    Zheng, and Zhiyong Yu

    [Yang et al., 2015] Dingqi Yang, Daqing Zhang, Vincent W. Zheng, and Zhiyong Yu. Modeling user activity prefer- ence by leveraging user spatial temporal characteristics in lbsns. IEEE Trans. Syst. Man Cybern. Syst. , 45(1):129– 142,

    2015

  14. [34]

    Next POI recommen- dation with dynamic graph and explicit dependency

    [Yin et al., 2023] Feiyu Yin, Yong Liu, Zhiqi Shen, Lisi Chen, Shuo Shang, and Peng Han. Next POI recommen- dation with dynamic graph and explicit dependency. In AAAI, pages 4827–4834. AAAI Press,

    2023

  15. [35]

    Next point-of- interest recommendation with inferring multi-step future preferences

    [Zhang et al., 2022] Lu Zhang, Zhu Sun, Ziqing Wu, Jie Zhang, Yew Soon Ong, and Xinghua Qu. Next point-of- interest recommendation with inferring multi-step future preferences. In IJCAI, pages 3751–3757. ijcai.org,

    2022

  16. [36]

    Collm: In- tegrating collaborative embeddings into large language models for recommendation

    [Zhang et al., 2023] Yang Zhang, Fuli Feng, Jizhi Zhang, Keqin Bao, Qifan Wang, and Xiangnan He. Collm: In- tegrating collaborative embeddings into large language models for recommendation. CoRR, abs/2310.19488,

    2023 arXiv

  17. [37]

    Exploring urban semantics: A multimodal model for POI semantic annota- tion with street view images and place names

    [Zhang et al., 2024] Dabin Zhang, Meng Chen, Weiming Huang, Yongshun Gong, and Kai Zhao. Exploring urban semantics: A multimodal model for POI semantic annota- tion with street view images and place names. In IJCAI, pages 2533–2541. ijcai.org,

    2024

  18. [38]

    Sheng, and Xiaofang Zhou

    [Zhao et al., 2022] Pengpeng Zhao, Anjing Luo, Yanchi Liu, Jiajie Xu, Zhixu Li, Fuzhen Zhuang, Victor S. Sheng, and Xiaofang Zhou. Where to go next: A spatio-temporal gated network for next POI recommendation. IEEE Trans. Knowl. Data Eng., 34(5):2512–2524, 2022

    2022

  19. [1997]

    Learning time slot preferences via mobility tree for next POI recommenda- tion

    [Huang et al., 2024] Tianhao Huang, Xuan Pan, Xiangrui Cai, Ying Zhang, and Xiaojie Yuan. Learning time slot preferences via mobility tree for next POI recommenda- tion. In Thirty-Eighth AAAI Conference on Artificial Intel- ligence, pages 8535–8543. AAAI Press,

    2024

  20. [2010]

    GPT4GEO: how a language model sees the world’s geography

    [Roberts et al., 2023] Jonathan Roberts, Timo L ¨uddecke, Sowmen Das, Kai Han, and Samuel Albanie. GPT4GEO: how a language model sees the world’s geography. CoRR, abs/2306.00020,

    2023 arXiv

  21. [2011]

    External eval- uation of ranking models under extreme position-bias

    [Fairstein et al., 2022] Yaron Fairstein, Elad Haramaty, Arnon Lazerson, and Liane Lewin-Eytan. External eval- uation of ranking models under extreme position-bias. In WSDM ’22: The Fifteenth ACM International Conference on Web Search and Data Mining , pages 252–261. ACM,

    2022

  22. [2013]

    Myers, and Jure Leskovec

    [Cho et al., 2011] Eunjoon Cho, Seth A. Myers, and Jure Leskovec. Friendship and mobility: user movement in location-based social networks. In Proceedings of the 17th ACM SIGKDD International Conference on Knowledge Discovery and Data Mining , pages 1082–1090. ACM,

    2011

  23. [2015]

    Getnext: Trajectory flow map enhanced transformer for next POI recommendation

    [Yang et al., 2022] Song Yang, Jiamou Liu, and Kaiqi Zhao. Getnext: Trajectory flow map enhanced transformer for next POI recommendation. In SIGIR, pages 1144–1153. ACM,

    2022

  24. [2016]

    Long short-term memory

    [Hochreiter and Schmidhuber, 1997] Sepp Hochreiter and J¨urgen Schmidhuber. Long short-term memory. Neural Comput., 9(8):1735–1780,

    1997

  25. [2018]

    HST- LSTM: A hierarchical spatial-temporal long-short term memory network for location prediction

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

    2018

  26. [2020]

    City foundation models for learning general purpose representations from openstreetmap

    [Balsebre et al., 2024] Pasquale Balsebre, Weiming Huang, Gao Cong, and Yi Li. City foundation models for learning general purpose representations from openstreetmap. In CIKM, pages 87–97. ACM,

    2024

  27. [2021]

    E4srec: An ele- gant effective efficient extensible solution of large lan- guage models for sequential recommendation

    [Li et al., 2023] Xinhang Li, Chong Chen, Xiangyu Zhao, Yong Zhang, and Chunxiao Xing. E4srec: An ele- gant effective efficient extensible solution of large lan- guage models for sequential recommendation. CoRR, abs/2312.02443,

    2023 arXiv

  28. [2022]

    Person- alized ranking metric embedding for next new POI rec- ommendation

    [Feng et al., 2015] Shanshan Feng, Xutao Li, Yifeng Zeng, Gao Cong, Yeow Meng Chee, and Quan Yuan. Person- alized ranking metric embedding for next new POI rec- ommendation. In IJCAI, pages 2069–2075. AAAI Press,

    2015

  29. [2023]

    Where to go next: Modeling long- and short-term user pref- erences for point-of-interest recommendation

    [Sun et al., 2020] Ke Sun, Tieyun Qian, Tong Chen, Yile Liang, Quoc Viet Hung Nguyen, and Hongzhi Yin. Where to go next: Modeling long- and short-term user pref- erences for point-of-interest recommendation. In AAAI, pages 214–221. AAAI Press,

    2020

  30. [2024]

    Lon- glora: Efficient fine-tuning of long-context large language models

    [Chen et al., 2024] Yukang Chen, Shengju Qian, Haotian Tang, Xin Lai, Zhijian Liu, Song Han, and Jiaya Jia. Lon- glora: Efficient fine-tuning of long-context large language models. In ICLR. OpenReview.net,

    2024

Pith tools

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