Pith. sign in

REVIEW 3 major objections 6 minor 2 cited by

Generative Next POI Recommendation with Semantic ID

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Replacing random POI IDs with semantic prefix-sharing codes improves LLM next-visit prediction by up to 16%.

desk verdict Solid extension of semantic-ID generative retrieval to next-POI, but the headline gains over LLM4POI are confounded by an unablated data augmentation. read the letter →

arxiv 2506.01375 v2 pith:WNKB5I55 submitted 2025-06-02 cs.IR

Dongsheng Wang , Yuxi Huang , Shen Gao , Yifan Wang , Chengrui Huang , Shuo Shang This is my paper
classification cs.IR
keywords nextPOIrecommendationsemanticIDgenerativeLLMfine-tuningresidualquantizedvariationalautoencoderdiversitylosscodebookquantizationout-of-domaingeneralization
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 an LLM can predict the next place a user will visit—the next point of interest (POI)—far better if each place is given a meaningful semantic ID instead of a random number. It proposes GNPR-SID, which first converts each POI's category, region, visiting-time pattern, and frequent visitors into a short hierarchical code, so similar places share code prefixes, and then fine-tunes a large language model to generate the next code from a user's check-in history. On check-in data from New York, Tokyo, and California, the method reports top-1 accuracy of 0.3618, 0.3062, and 0.2403, beating the strongest random-ID LLM baseline by 7%, 1%, and 16% respectively. The point of the design is that the ID itself carries semantic and collaborative information, so the LLM can reason about places instead of treating each ID as an opaque token. A reader should care because the same trick—replacing opaque tokens with structured, semantics-carrying symbols—may transfer to other generative retrieval tasks.

What carries the argument

The load-bearing mechanism is the semantic ID (SID) viewed as a hierarchical code: a short tuple of codeword indices produced by residual quantization, in which the first index gives the coarsest semantic bucket and each later index refines it. The construction module is a residual quantized variational autoencoder (RQVAE): an encoder maps the concatenated POI features to a latent vector, each codebook layer quantizes the residual left by the previous layer, the selected codewords are summed and decoded to reconstruct the features, and the index sequence becomes the SID. A diversity loss balances two constraints—codewords should be used roughly uniformly across the codebook, and vectors assigned to the same codeword should stay close—so the discrete space does not collapse. The generative module then treats each codeword index as a token, so the LLM is trained to generate the next POI by completing the SID sequence with timestamps.

What would settle it

Retrain GNPR-SID on NYC with the learned codebook but with the codeword indices randomly permuted within every codebook level, so each POI keeps its quantized feature vector yet its SID string loses prefix semantics; if top-1 accuracy remains near 0.3618 rather than dropping toward the random-ID baseline's 0.3372, the reported gain is not caused by semantic prefix sharing.

Watch

Extended reading notes

Core claim

The central claim, stated the way the authors would state it, is that semantic IDs make generative next-POI recommendation accurate and portable in a way random IDs cannot. Each POI is described by a one-hot concatenation of category, grid region, top frequent time slots, and top frequent visitors; an RQVAE encoder maps this to a latent vector, and multiple layers of codebooks quantize the residuals so that the POI's SID is a tuple of codeword indices. The design deliberately lets similar POIs share early indices: a tea shop near a coffee shop will receive the same leading codewords and differ only in a later refinement, and any remaining same-ID collisions are resolved by appending a unique symbol. A large language model is fine-tuned with low-rank adapters on prompts that interleave the user's historical SIDs with their visit timestamps, and the next POI is the next SID the model generates. The paper reports Acc@1 of 0.3618, 0.3062, and 0.2403 on NYC, TKY, and CA, and its ablations show that swapping SIDs for random IDs, dropping timestamps, removing the diversity loss, or removing the quantization loss all lower accuracy, while removing the reconstruction loss prevents training from converging.

Load-bearing premise

The load-bearing premise is that a semantic codebook and the LLM's SID vocabulary trained on one city's POIs can be applied to another city's POIs and still yield semantically meaningful IDs; the paper does not specify how target-city POIs are quantized with a source codebook, how features are normalized, or how the LLM can emit codewords it never saw for the target domain.

Editorial extensions

If this is right

  • On the three datasets, replacing random numeric IDs with SIDs raises top-1 accuracy by 7%, 1%, and 16% over the strongest random-ID LLM baseline, and removing the SID in ablation lowers accuracy on every dataset.
  • A model fine-tuned on one city transfers to another city with less accuracy loss than the random-ID baseline, as shown by the cross-dataset rows of the comparison table.
  • Because all POI metadata is stored in a few code tokens, the SID format cuts total prompt tokens, training time, and test time by roughly half compared to the random-ID baseline on the NYC benchmark.
  • The diversity loss has a usable operating range: too little produces codeword collisions and drops accuracy, while too much forces dissimilar POIs together and also drops accuracy.
  • The SID prefixes align with POI semantics: visualization on TKY shows different first-codeword values correspond to distinct venue categories such as transportation versus entertainment.

Reading between the lines

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

  • An extension not tested in the paper: the out-of-domain experiment trains on one city and tests on another, so it mixes two transfer questions—whether the codebook's quantization geometry transfers, and whether the LLM's learned mapping from SIDs to POIs transfers; retraining one component while freezing the other would separate them.
  • A testable extension of the paper's motivation: keep a held-out set of truly unseen POIs, since the current protocol deletes test POIs absent from the training set, to measure cold-start generalization directly.
  • A design option the paper leaves implicit: predict only the first codeword (region or category), filter candidate POIs, then decode the remaining indices—a coarse-to-fine decoding that would shrink the output vocabulary at each step.
  • A general transfer of the recipe: quantize any entity's features into prefix-sharing tokens and fine-tune an LLM to emit them, which could apply to products, documents, or other generative retrieval settings where entities are currently assigned random IDs.
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 / 6 minor

Summary. The paper proposes GNPR-SID, an LLM-based next-POI recommender in which each POI is represented by a semantic ID (SID). The SID is produced by an RQVAE quantizer over a feature vector that combines POI category, Plus-Code region, frequent time slots, and frequent-visitor collaborative signals. An LLM (LLaMA3-8B) is fine-tuned with LoRA to generate the next POI's SID from a history of SIDs and timestamps. Experiments on Foursquare-NYC, Foursquare-TKY, and Gowalla-CA report Acc@1 improvements over LLM4POI, ablations of SID, time, diversity loss, quantization loss, and reconstruction loss, an analysis of diversity loss, a visualization of SID semantic clusters, an out-of-domain transfer experiment, and an efficiency comparison.

Significance. If the empirical claims are robust, the paper makes a useful contribution by adapting semantic ID tokenization, already successful in generative retrieval, to POI recommendation: the two-module design (semantic ID construction plus generative LLM fine-tuning) is clean, the code is released, and the ablation structure (w/o SID, w/o Time, w/o Ldiv, w/o Lquant) is a strength. However, the central 'up to 16% improvement' claim is currently confounded by an unablated data-augmentation difference between GNPR-SID and the baselines, and the cross-dataset generalization evidence is underspecified. The paper is likely to interest the KDD community once these issues are resolved, but the headline quantitative conclusion is not yet supported.

major comments (3)
  1. [Appendix A.1.2 / Section 5.3 / Table 2] The fill-in-the-blank data augmentation described in Appendix A.1.2 (20% of cropped samples are replaced by a historical-position prediction target) is used when constructing the GNPR-SID training set, but Section 5.3 does not state that LLM4POI or any other baseline receives the same augmentation, and no ablation removes this augmentation. The w/o SID row in Table 3 keeps the augmentation and is therefore not a control for it; on CA that random-ID version already reaches 0.2296 versus LLM4POI's 0.2065, i.e., about 11% relative improvement without any semantic ID. As a result, the Table 2 margins, including the advertised 16% improvement on CA, cannot be attributed to SIDs. Please add an ablation that turns the fill-in-the-blank augmentation on and off, and run the main comparison under matched training conditions (e.g., GNPR-SID without augmentation, and LLM4POI with the same augmentation).
  2. [Tables 2-4 / Section 6.1] All reported Acc@1 values are single point estimates with no error bars, confidence intervals, or significance tests. The TKY comparison is especially fragile: GNPR-SID achieves 0.3062 versus LLM4POI's 0.3035, a relative difference of about 0.9% that could easily be within run-to-run noise. The claim that GNPR-SID 'consistently outperforms' the state of the art needs variance information at least for the main results and for the w/o SID ablation. Please report multiple seeds or bootstrap confidence intervals, with a paired significance test where appropriate.
  3. [Section 5.1 / Table 5 / Section 6.5] The paper's introduction claims that SIDs improve generalization to newly added, unseen POIs, but Section 5.1 removes from the test set all users and POIs not present in the training set, so no within-dataset test POI is genuinely unseen. The only out-of-domain evidence is Table 5, yet the protocol for applying a source-trained RQVAE codebook to target-dataset POIs is not specified: it is unclear whether target features are normalized with source statistics, how the source codebook is reused, how feature-dimension mismatches across cities (different category sets, user sets) are handled, and whether the target test set also excludes POIs absent from the target training split. Please provide the exact cross-dataset encoding procedure and, if possible, add an evaluation setting that contains genuinely new POIs in the test set.
minor comments (6)
  1. [Section 5.2 / References] The citations for LLaMA3-8B and LoRA appear swapped: the text cites LLaMA3-8B as [15] and LoRA as [7], but in the reference list [15] is the LoRA paper and [7] is the Llama 3 report.
  2. [Section 2] The phrase 'it achieves the stats-of-the-art performance' should be 'state-of-the-art'.
  3. [Equation (13)] The compactness loss notation is confusing: the denominator is k(k-1) and the sum is over codeword vectors, but k is also used for the number of codewords in a layer; please clarify the indexing and the definition of the summation range.
  4. [Figure 3] Because SIDs are constructed from category, region, time-slot, and user features, the observation that similar POIs share prefixes is partly a sanity check of the construction rather than an independent validation; consider adding a quantitative evaluation, such as average prefix overlap against category or geographic similarity, on held-out POIs.
  5. [Section 6.6 / Table 6] The efficiency comparison would be more informative if the paper stated whether LLM4POI was re-implemented with the same tokenizer, generation settings, and padding, or whether the numbers were taken from the original paper; otherwise the token and time reductions may reflect implementation choices rather than the proposed method.
  6. [Appendix A.1.1] The multiple-cropping rule is stated vaguely: 'exceeds several times the given length' is not precise. Please specify the exact segmentation and cropping thresholds used in the experiments.

Circularity Check

1 steps flagged · score 2.0 of 10

Central accuracy claim is evaluated on held-out test data and is not circular; only the SID semantic-similarity validation reduces to the design objective by construction.

  1. self definitional [Section 6.4 (Semantic Relevance Analysis of SID), with Eq. (1), (8), (9)]
    "In Figure 3a, we randomly select four SID prefixes with varying first codeword, and show the distribution of these SIDs across the corresponding POI categories. From Figure 3a we can find that different semantic spaces (a.k.a., different SID prefixes) effectively capture information related to distinct semantic categories of POIs, while POIs of similar categories are quantized into the same semantic space (a.k.a., the same SID prefix)."

    The category c is an explicit input feature of p_e in Eq. (1), and the SID is the tuple of codebook indices (Eq. (8)) trained with reconstruction loss Lrecon = ||p_e - p_e_hat||^2 (Eq. (9)). Because the autoencoder must reconstruct the one-hot category to minimize Lrecon, the SID is forced to encode category information. Showing that SID prefixes align with categories therefore confirms the reconstruction objective by construction rather than providing independent evidence that SIDs capture semantics. This is a sanity check of the design, not a load-bearing derivation for the main accuracy claim.

full rationale

The main derivation chain—constructing SIDs from POI features via RQVAE and fine-tuning LLaMA-3 to generate the next SID—is self-contained and tested on held-out next-POI labels. The Table 2 gains and the w/o SID ablation in Table 3 compare against baselines and randomized-ID variants on the same evaluation protocol, so the central accuracy claim does not reduce to the input features or to a fitted parameter. I find no self-citation that is load-bearing: the cited RQVAE [47], LoRA [15], and prompt-construction works are external or used as standard tools, not as a uniqueness theorem. The only circularity I identified is auxiliary: the Section 6.4 semantic-relevance analysis verifies that SID prefixes align with POI categories, but category is an explicit input feature in Eq. (1) and the reconstruction loss (Eq. (9)) forces the SID to encode it; hence Figure 3 mostly demonstrates that the autoencoder trained correctly, not an independent discovery. This does not affect the main accuracy result. The Appendix A.1 data-augmentation protocol is not stated to be applied to baselines, which is a potential experimental confound, but that is a fairness/correctness concern rather than a circularity of the derivation, and per the review rules it is not scored here.

Assumptions & free parameters 7 free parameters · 4 assumptions · 0 invented entities

The performance of GNPR-SID rests on hand-chosen hyperparameters and domain assumptions rather than on derived constants. The SID construction is learned, not derived, and the feature vector is a hand-assembled concatenation. The central empirical claim, however, is evaluated on held-out data, so these parameters are fit to validation performance, not to the test labels.

free parameters (7)
  • Diversity loss weight lambda = 0.25
    Chosen in Table 4 as the best Acc@1 on NYC/TKY; the central claim depends on the SID quality this loss controls.
  • Commitment loss weight beta = not reported
    Appears in Eq. 10; controls encoder commitment but its value is not stated in the implementation details.
  • Number of codebook layers L = 3
    Sets SID length; all experiments use 3 layers.
  • Codebook size K per layer = 32 (NYC), 64 (TKY, CA)
    Determines semantic granularity of the SID space.
  • Codeword dimension d = 64
    Embedding dimension of codeword vectors.
  • Top-10 time slots = 10
    Number of most frequent time slots used in the temporal signal t.
  • Top-10 visiting users = 10
    Number of most frequent visitors used in the collaborative signal c_u.
assumptions (4)
  • domain assumption Residual quantized VAE assigns semantically similar POIs to shared SID prefixes (Eqs. 4-8)
    Not proven; the paper validates this empirically in §6.4, but it is a core premise of the method.
  • domain assumption The concatenated feature vector p_e (Eq. 1) captures the semantic essence of a POI
    Hand-crafted feature selection; if these features are insufficient, the SIDs carry little semantics and the approach loses its advantage.
  • domain assumption Fine-tuning LLaMA-3 with LoRA can learn the mapping from historical SID sequences to the next SID (Section 4.2)
    Depends on the LLM's ability to internalize discrete codes and generalize to new combinations.
  • standard math Stop-gradient and standard VAE optimization (Eq. 10) behave as in the VQ-VAE literature
    Borrowed from Van Den Oord et al. 2017; the paper relies on established practice.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generative Next POI Recommendation with Semantic ID." pith.science (2026). https://pith.science/paper/WNKB5I55

@misc{pith2026250601375,
  author       = {Pith},
  title        = {Pith review of: Generative Next POI Recommendation with Semantic ID},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WNKB5I55}},
  note         = {Machine review of arXiv:2506.01375}
}
read the original abstract

Point-of-interest (POI) recommendation systems aim to predict the next destinations of user based on their preferences and historical check-ins. Existing generative POI recommendation methods usually employ random numeric IDs for POIs, limiting the ability to model semantic relationships between similar locations. In this paper, we propose Generative Next POI Recommendation with Semantic ID (GNPR-SID), an LLM-based POI recommendation model with a novel semantic POI ID (SID) representation method that enhances the semantic understanding of POI modeling. There are two key components in our GNPR-SID: (1) a Semantic ID Construction module that generates semantically rich POI IDs based on semantic and collaborative features, and (2) a Generative POI Recommendation module that fine-tunes LLMs to predict the next POI using these semantic IDs. By incorporating user interaction patterns and POI semantic features into the semantic ID generation, our method improves the recommendation accuracy and generalization of the model. To construct semantically related SIDs, we propose a POI quantization method based on residual quantized variational autoencoder, which maps POIs into a discrete semantic space. We also propose a diversity loss to ensure that SIDs are uniformly distributed across the semantic space. Extensive experiments on three benchmark datasets demonstrate that GNPR-SID substantially outperforms state-of-the-art methods, achieving up to 16% improvement in recommendation accuracy.

Figures

Figures reproduced from arXiv: 2506.01375 by the authors.

Figure 1
Figure 1. Comparison of random ID and semantic ID: Seman [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The overview of the Generative Next POI Recommendation with Semantic ID (GNPR-SID). It consists of two modules: [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. SIDs semantic relevance analysis. to different semantic clusters, thereby negatively impacting the performance of the next POI recommendation. 6.4 Semantic Relevance Analysis of SID To intuitively validate whether the SIDs generated by our proposed method can effectively model the semantic information of POIs, we visualize the POIs in the TKY dataset. In Figure 3a, we randomly select four SID prefixes with varying f… view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Replace the next POI recommendation task with a [PITH_FULL_IMAGE:figures/full_fig_p011_4.png]

Discussion (0). Sign in 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. Guess Where You Go: Generative Next Point-of-Interest Recommendation in Amap

    cs.IR 2026-07 conditional novelty 5.0 of 10

    Gwhere generates the next POI as a token sequence of contrastively-learned semantic IDs and reports production CTR gains of 5.83% and 6.20% in Amap.

  2. 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

54 extracted references · 32 canonical work pages · cited by 2 Pith papers

  1. [1]

    Jinze Bai, Shuai Bai, Yunfei Chu, Zeyu Cui, Kai Dang, Xiaodong Deng, Yang Fan, Wenbin Ge, Yu Han, Fei Huang, et al. 2023. Qwen technical report. arXiv preprint arXiv:2309.16609 (2023)

    arXiv 2023

  2. [2]

    Jinze Bai, Shuai Bai, Shusheng Yang, Shijie Wang, Sinan Tan, Peng Wang, Junyang Lin, Chang Zhou, and Jingren Zhou. 2023. Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond.arXiv preprint arXiv:2308.12966 (2023)

    arXiv 2023

  3. [3]

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Neelakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. 2020. Language models are few-shot learners. Advances in neural information processing systems 33 (2020), 1877–1901

    2020

  4. [4]

    Chen Cheng, Haiqin Yang, Michael R Lyu, and Irwin King. 2013. Where you like to go next: Successive point-of-interest recommendation. In Twenty-Third international joint conference on Artificial Intelligence

    work page 2013

  5. [5]

    Eunjoon Cho, Seth A Myers, and Jure Leskovec. 2011. 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 . 1082–1090

    work page 2011

  6. [6]

    Chenghua Duan, Wei Fan, Wei Zhou, Hu Liu, and Junhao Wen. 2023. Clsprec: Contrastive learning of long and short-term preferences for next poi recommen- dation. In Proceedings of the 32nd acm international conference on information and knowledge management. 473–482

    work page 2023

  7. [7]

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

    arXiv 2024

  8. [8]

    Shanshan Feng, Xutao Li, Yifeng Zeng, Gao Cong, Yeow Meng Chee, and Quan Yuan. 2015. Personalized ranking metric embedding for next new poi recommen- dation. (2015). KDD ’25, August 3–7, 2025, Toronto, ON, Canada Dongsheng Wang et al

    work page 2015

Show all 54 references
  1. [9]

    Shanshan Feng, Haoming Lyu, Fan Li, Zhu Sun, and Caishun Chen. 2024. Where to move next: Zero-shot generalization of llms for next poi recommendation. In 2024 IEEE Conference on Artificial Intelligence (CAI) . IEEE, 1530–1535

    2024

  2. [10]

    Shanshan Feng, Feiyu Meng, Lisi Chen, Shuo Shang, and Yew Soon Ong. 2024. Rotan: A rotation-based temporal attention network for time-specific next poi rec- ommendation. In Proceedings of the 30th ACM SIGKDD Conference on Knowledge Discovery and Data Mining . 759–770

    2024

  3. [11]

    Shijie Geng, Shuchang Liu, Zuohui Fu, Yingqiang Ge, and Yongfeng Zhang. 2022. Recommendation as Language Processing (RLP): A Unified Pretrain, Personalized Prompt & Predict Paradigm (P5). In Proceedings of the 16th ACM Conference on Recommender Systems. 299–315

    2022

  4. [12]

    Alex Graves and Alex Graves. 2012. Long short-term memory.Supervised sequence labelling with recurrent neural networks (2012), 37–45

    2012

  5. [13]

    Jiaming Han, Kaixiong Gong, Yiyuan Zhang, Jiaqi Wang, Kaipeng Zhang, Dahua Lin, Yu Qiao, Peng Gao, and Xiangyu Yue. 2024. OneLLM: One Framework to Align All Modalities with Language. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) . ...

    2024

  6. [14]

    Jing He, Xin Li, Lejian Liao, Dandan Song, and William Cheung. 2016. Inferring a personalized next point-of-interest recommendation model with latent behavior patterns. In Proceedings of the AAAI conference on artificial intelligence , Vol. 30

    2016

  7. [15]

    Edward J Hu, Yelong Shen, Phillip Wallis, Zeyuan Allen-Zhu, Yuanzhi Li, Shean Wang, Lu Wang, and Weizhu Chen. 2021. Lora: Low-rank adaptation of large language models. arXiv preprint arXiv:2106.09685 (2021)

    2021 arXiv

  8. [16]

    Md Ashraful Islam, Mir Mahathir Mohammad, Sarkar Snigdha Sarathi Das, and Mohammed Eunus Ali. 2022. A survey on deep learning based Point-of-Interest (POI) recommendations. Neurocomputing 472 (2022), 306–325

    2022

  9. [17]

    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

  10. [18]

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

  11. [19]

    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

  12. [20]

    Nicholas Lim, Bryan Hooi, See-Kiong Ng, Yong Liang Goh, Renrong Weng, and Rui Tan. 2022. Hierarchical multi-task graph recurrent network for next poi recommendation. In Proceedings of the 45th international ACM SIGIR conference on Research and development in Information Retrie...

    2022

  13. [21]

    Nicholas Lim, Bryan Hooi, See-Kiong Ng, Xueou Wang, Yong Liang Goh, Ren- rong Weng, and Jagannadan Varadarajan. 2020. STP-UDGAT: Spatial-temporal- preference user dimensional graph attention network for next POI recommenda- tion. In Proceedings of the 29th ACM International co...

    2020

  14. [22]

    Qiang Liu, Shu Wu, Liang Wang, and Tieniu Tan. 2016. Predicting the next location: A recurrent model with spatial and temporal contexts. In Proceedings of the AAAI conference on artificial intelligence , Vol. 30

    2016

  15. [23]

    Shuai Liu, Ning Cao, Yile Chen, Yue Jiang, and Gao Cong. 2024. nextlocllm: next location prediction using LLMs. arXiv preprint arXiv:2410.09129 (2024)

    2024

  16. [24]

    Yiding Liu, Tuan-Anh Nguyen Pham, Gao Cong, and Quan Yuan. 2017. An experimental evaluation of point-of-interest recommendation in location-based social networks. (2017)

    2017

  17. [25]

    Yan Luo, Haoyi Duan, Ye Liu, and Fu-Lai Chung. 2023. Timestamps as Prompts for Geography-Aware Location Recommendation. InProceedings of the 32nd ACM International Conference on Information and Knowledge Management . 1697–1706

    2023

  18. [26]

    Yingtao Luo, Qiang Liu, and Zhaocheng Liu. 2021. Stan: Spatio-temporal attention network for next location recommendation. In Proceedings of the web conference

    2021

  19. [27]

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

  20. [28]

    Pablo Sánchez and Alejandro Bellogín. 2022. Point-of-interest recommender systems based on location-based social networks: a survey from an experimental perspective. ACM Computing Surveys (CSUR) 54, 11s (2022), 1–37

    2022

  21. [29]

    Ke Sun, Tieyun Qian, Tong Chen, Yile Liang, Quoc Viet Hung Nguyen, and Hongzhi Yin. 2020. Where to go next: Modeling long-and short-term user prefer- ences for point-of-interest recommendation. InProceedings of the AAAI conference on artificial intelligence, Vol. 34. 214–221

    2020

  22. [30]

    Weiwei Sun, Lingyong Yan, Zheng Chen, Shuaiqiang Wang, Haichao Zhu, Pengjie Ren, Zhumin Chen, Dawei Yin, Maarten Rijke, and Zhaochun Ren. 2024. Learning to tokenize for generative retrieval. Advances in Neural Information Processing Systems 36 (2024)

    2024

  23. [31]

    Aaron Van Den Oord, Oriol Vinyals, et al. 2017. Neural discrete representation learning. Advances in neural information processing systems 30 (2017)

    2017

  24. [32]

    En Wang, Yiheng Jiang, Yuanbo Xu, Liang Wang, and Yongjian Yang. 2022. Spatial- temporal interval aware sequential POI recommendation. In 2022 IEEE 38th international conference on data engineering (ICDE) . IEEE, 2086–2098

    2022

  25. [33]

    Jiawei Wang, Renhe Jiang, Chuang Yang, Zengqing Wu, Makoto Onizuka, Ryosuke Shibasaki, Noboru Koshizuka, and Chuan Xiao. 2024. Large language models as urban residents: An llm agent framework for personal mobility gener- ation. arXiv preprint arXiv:2402.14744 (2024)

    2024 arXiv

  26. [34]

    Wenjie Wang, Honghui Bao, Xinyu Lin, Jizhi Zhang, Yongqi Li, Fuli Feng, See- Kiong Ng, and Tat-Seng Chua. 2024. Learnable item tokenization for generative recommendation. In Proceedings of the 33rd ACM International Conference on Information and Knowledge Management . 2400–2409

    2024

  27. [35]

    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

  28. [36]

    Yujing Wang, Yingyan Hou, Haonan Wang, Ziming Miao, Shibin Wu, Qi Chen, Yuqing Xia, Chengmin Chi, Guoshuai Zhao, Zheng Liu, et al . 2022. A neural corpus indexer for document retrieval. Advances in Neural Information Processing Systems 35 (2022), 25600–25614

    2022

  29. [37]

    Yidan Wang, Zhaochun Ren, Weiwei Sun, Jiyuan Yang, Zhixiang Liang, Xin Chen, Ruobing Xie, Su Yan, Xu Zhang, Pengjie Ren, et al . 2024. Enhanced generative recommendation via content and collaboration integration. arXiv preprint arXiv:2403.18480 (2024)

    2024 arXiv

  30. [38]

    Zhaobo Wang, Yanmin Zhu, Haobing Liu, and Chunyang Wang. 2022. Learn- ing graph-based disentangled representations for next POI recommendation. In Proceedings of the 45th international ACM SIGIR conference on research and development in information retrieval. 1154–1163

    2022

  31. [39]

    Zhaobo Wang, Yanmin Zhu, Chunyang Wang, Wenze Ma, Bo Li, and Jiadi Yu

  32. [40]

    Wilson Wongso, Hao Xue, and Flora D Salim. 2024. GenUP: Generative User Profilers as In-Context Learners for Next POI Recommender Systems. arXiv preprint arXiv:2410.20643 (2024)

    2024 arXiv

  33. [41]

    Yuxia Wu, Ke Li, Guoshuai Zhao, and Xueming Qian. 2020. Personalized long-and short-term preference learning for next POI recommendation. IEEE Transactions on Knowledge and Data Engineering 34, 4 (2020), 1944–1957

    2020

  34. [42]

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

    Adaptive Graph Representation Learning for Next POI Recommendation. In Proceedings of the 46th International ACM SIGIR Conference on Research and Development in Information Retrieval . 393–402

  35. [43]

    Xiaodong Yan, Tengwei Song, Yifeng Jiao, Jianshan He, Jiaotuan Wang, Ruopeng Li, and Wei Chu. 2023. Spatio-temporal hypergraph learning for next POI rec- ommendation. In Proceedings of the 46th international ACM SIGIR conference on research and development in information retri...

    2023

  36. [44]

    Dingqi Yang, Daqing Zhang, Vincent W Zheng, and Zhiyong Yu. 2014. Modeling user activity preference by leveraging user spatial temporal characteristics in LBSNs. IEEE Transactions on Systems, Man, and Cybernetics: Systems 45, 1 (2014), 129–142

    2014

  37. [45]

    Hao Xue, Bhanu Prakash Voutharoja, and Flora D. Salim. 2022. Leveraging language foundation models for human mobility forecasting. In Proceedings of the 30th International Conference on Advances in Geographic Information Systems . 9

    2022

  38. [46]

    Shukang Yin, Chaoyou Fu, Sirui Zhao, Ke Li, Xing Sun, and Chen Enhong Xu Tong. 2023. A Survey on Multimodal Large Language Models. arXiv preprint arXiv:2306.13549 (2023)

    2023 arXiv

  39. [47]

    Neil Zeghidour, Alejandro Luebs, Ahmed Omran, Jan Skoglund, and Marco Tagliasacchi. 2021. Soundstream: An end-to-end neural audio codec. IEEE/ACM Transactions on Audio, Speech, and Language Processing 30 (2021), 495–507

    2021

  40. [48]

    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. 1144–1153

    2022

  41. [49]

    Pengpeng Zhao, Anjing Luo, Yanchi Liu, Jiajie Xu, Zhixu Li, Fuzhen Zhuang, Victor S Sheng, and Xiaofang Zhou. 2020. Where to go next: A spatio-temporal gated network for next poi recommendation. IEEE Transactions on Knowledge and Data Engineering 34, 5 (2020), 2512–2524

    2020

  42. [50]

    Bowen Zheng, Yupeng Hou, Hongyu Lu, Yu Chen, Wayne Xin Zhao, Ming Chen, and Ji-Rong Wen. 2024. Adapting Large Language Models by Integrating Collab- orative Semantics for Recommendation. In IEEE 40th International Conference on Data Engineering (ICDE). 1435–1448. Generative Ne...

    2024

  43. [51]

    Lu Zhang, Zhu Sun, Ziqing Wu, Jie Zhang, Yew Soon Ong, and Xinghua Qu

  44. [2022]

    In IJCAI

    Next Point-of-Interest Recommendation with Inferring Multi-step Future Preferences.. In IJCAI. 3751–3757

  45. [2023]

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

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

    2023

  46. [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 7, 2026 · model on record in the stance chip above.