Pith. sign in

REVIEW 3 major objections 7 minor 47 references

TrajGEOS: Trajectory Graph Enhanced Orientation-based Sequential Network for Mobility Prediction

T0 review · 3 major / 7 minor · reviewed 2026-08-11 · 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 TrajGEOS, a trajectory graph enhanced orientation-based sequential network, outperforms existing state-of-the-art baselines on next location prediction across three real-world LBSN datasets.

desk verdict TrajGEOS is a competent architecture with honest ablations, but the state-of-the-art claim is not established because baselines run on non-identical data and hyperparameters are tuned on the NYC test set. read the letter →

arxiv 2412.19092 v1 pith:R26QBXPW submitted 2024-12-26 cs.AI

Zhaoping Hu , Zongyuan Huang , Jinming Yang , Tao Yang , Yaohui Jin , Yanyan Xu This is my paper · ORCID
classification cs.AI
keywords nextlocationpredictionhumanmobilitytrajectorygraphneuralnetworksattentionmechanismmulti-tasklearninglocation-basedsocialsequentialrecommendation
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 sets out to improve next-location prediction, the task of guessing which place a person will check in next, by exploiting structure hidden in everyone's movement histories. It argues that standard sequential models overlook the relationships between locations and the pull of a user's most recent weeks on their next move. TrajGEOS builds a global trajectory graph from all users' check-in sequences, learns location and user embeddings through hierarchical graph convolution, and adds an orientation module that attends over the past two to three weeks of a user's trajectory to form a mid-term preference. On three public location-based social network datasets, the model reports higher Recall@1 than the strongest baselines, and ablations credit the gain mainly to the graph embeddings and the mid-term preference signal.

What carries the argument

The central object is the trajectory graph, a directed graph whose nodes are locations and whose edges are observed transitions from all users' training histories, each edge annotated with distance, total transition count, and a 24-dimensional hourly flow vector. Hierarchical graph convolution applies an edge-aware message-passing layer on this global graph to produce location embeddings, then applies a second convolution on each user's subgraph and performs a visit-frequency-weighted readout to obtain a long-term user embedding. The orientation module then adds sinusoidal position encodings to the encoded recent weekly sub-trajectories and uses attention, keyed by the short-term recurrent output, to form a mid-term preference; the three preferences and the user embedding are concatenated and fed to two predictors for next location and next category.

What would settle it

Re-run every baseline and TrajGEOS on one identical user-level 80/20 split of the same filtered check-in data for NYC and Dallas, with identical test weeks and filtering rules, and check whether TrajGEOS still leads on Recall@1 and MRR@10 across five seeds; if the margin collapses, the reported outperformance is not established.

Watch

Extended reading notes

Core claim

On its own terms, the paper's central claim is that TrajGEOS achieves state-of-the-art next-location prediction on the NYC, TKY, and Dallas check-in datasets, with Recall@1 of 27.2% versus 26.4% for the best baseline on NYC, 24.9% versus 23.8% on Tokyo, and 13.2% versus 13.1% on Dallas, and consistent leads on Recall@5, Recall@10, and MRR@10. The claimed mechanism is a fusion of three preference scales: long-term preference read out from a user's trajectory subgraph with visit-frequency weights, short-term preference from the current week's recurrent encoding, and a mid-term preference produced by an attention-based orientation module over the preceding two to three weeks. These are combined with location embeddings learned from a global trajectory graph whose edges carry distance, transition counts, and 24-hour flow patterns, and with a next-category prediction auxiliary task that shares the same recurrent outputs. Ablation experiments show that removing the graph module or the orientation module lowers accuracy, while removing the short-term preference causes the largest drop.

Load-bearing premise

The ranking conclusion assumes the separately preprocessed datasets used for different baseline models are comparable enough to compare their scores, even though each model was run on its own subset of users, locations, and check-in records.

Editorial extensions

If this is right

  • If the reported gains hold on identical test splits, TrajGEOS is the current state of the art on the three standard LBSN benchmarks and a template for combining graph-learned location relations with multi-scale sequential preference.
  • The trajectory graph's edge features (distance, transition volume, hourly flow) can be injected directly into location embeddings, which suggests richer mobility counts belong in the representation rather than only in the loss.
  • The attention-based orientation module gives a mid-term preference signal at lower cost than a second recurrent network; replacing it with a GRU in the paper's own experiments is consistently worse.
  • The multi-task category head acts as a regularizer: the best location Recall@1 is found at alpha = 0.7, showing the auxiliary category loss can be balanced against the main task without degrading it.
  • The visual analyses indicate the model's advantage is spatially closer predictions and weaker for users with high location or category entropy, identifying where further gains are most needed.

Reading between the lines

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

  • A skeptical first check would be to run every model on exactly one shared test split (same filtered users, locations, and held-out weeks); the per-model dataset statistics in the paper's Table I suggest such a check could change the ranking.
  • Because the trajectory graph is built from training transitions only, the same hierarchical-convolution-plus-recent-attention design should apply to other sequential prediction problems where item transitions form a graph, such as web navigation or purchase sequences.
  • The 24-hour flow edge feature could be extended beyond week segmentation, for instance by conditioning the orientation module on time of day or adding hour prediction as a third task.
  • The authors say the implementation will be released after publication; if that release includes the exact preprocessed data split for each model, the comparability question can be settled directly.
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

3 major / 7 minor

Summary. This paper proposes TrajGEOS, a deep model for next-location prediction from LBSN check-in data. The architecture combines three components: (i) a hierarchical graph module that runs an edge-aware graph convolution (EGraphSAGE) on a global trajectory graph built from users' training trajectories, with node features from location ID, category, and coordinates and edge features from transition counts, distance, and 24-hour flow, followed by a GraphSAGE on per-user subgraphs whose visit-frequency-weighted readout is treated as the user's long-term preference; (ii) a sequential module that encodes the current weekly sub-trajectory with a shared GRU to produce a short-term preference; and (iii) an orientation module that adds position encodings to the preceding two-to-three weeks of records and attends over them, using the GRU output as query, to produce a mid-term preference. A multi-task head predicts the next location and the next category with loss L = alpha*Ll + (1-alpha)*Lc. Experiments on NYC, TKY, and Dallas (Gowalla) report Recall@1/5/10 and MRR@10 against FPMC-D/W, DeepMove, LSTPM, GeoSAN, GETNext, and MTNet, claim to outperform all baselines on all three datasets, and include ablations, visualizations, and sensitivity analyses.

Significance. If the performance claim held, TrajGEOS would be a solid incremental contribution to next-location prediction: it packages graph-derived location embeddings with informative edge attributes (distance, transition counts, 24-hour flow), a user-subgraph readout for long-term preference, and a lightweight attention-based mid-term module. The paper deserves credit for constructing the trajectory graph from training records only, for using a user-level chronological split, for reporting five-run averages of its own model, and for publishing sensitivity tables for its main hyperparameters; the ablation study supports the value of the graph module (e.g., TKY R@1 drops from 0.2490 to 0.2017 without it) and of the orientation module, and the implementation is promised for release.

major comments (3)
  1. [Section V-A, Table I] The assertion in Section V-A that the per-model datasets 'exhibit minimal differences, allowing for a meaningful comparison' is contradicted by Table I. On Dallas, TrajGEOS uses 1,357 users / 118,069 records / 5,428 locations, while LSTPM uses 954 / 103,664 / 5,366, GeoSAN uses 2,300 / 142,980 / 5,357, and DeepMove uses 1,193 / 93,911 / 5,346. On NYC, record counts range from 111,968 (DeepMove) to 138,229 (GeoSAN), with TrajGEOS at 131,874, and the user sets differ as well (e.g., LSTPM 1,019 users). Because each model's trajectories are segmented into weekly sub-trajectories and evaluated on the chronological last 20% per user, differing record subsets imply differing test instances, and differing location sets imply differing candidate sets. The reported advantages over the best baseline (0.86 pp on NYC, 1.02 pp on TKY, 0.13 pp on Dallas) are of the same order as these dataset discrepancies. To support the claim that TrajGEOS outperforms the baselines, the authors should run all models under a single shared preprocessing pipeline (including the same user set, record set, and candidate location set for the test portion) or demonstrate on a common subset of test instances that the ranking is unchanged.
  2. [Appendix C, Tables IV, V, VII, VIII] The hyperparameters alpha, the recent-trajectory length kappa, the number of EGraphSAGE layers, and the number of GraphSAGE layers are each selected by Recall@1 computed on the NYC dataset, and the selected configurations (alpha = 0.7, kappa = 2, depth 2 in both graph modules) reproduce exactly the NYC R@1 = 0.2721 reported in Table II. This means the reported NYC result, and by transfer the TKY and Dallas results, inherit the best of several configurations evaluated on the test partition. The sensitivity spreads are bounded (at most about 1.0 pp for alpha in Table IV and 0.5 pp for kappa in Table V), so the tuning advantage alone is not huge, but it is comparable to the headline margin over the best baseline on NYC (0.86 pp), and the protocol is still methodologically incorrect. In addition, Section III presents kappa = 2 as a fixed definition ('we set kappa = 2'), whereas Table V shows it is an empirically chosen value. The authors should select all hyperparameters on a validation split (e.g., the sub-trajectories immediately preceding the last 20% per user) and report test-set results for the single chosen configuration.
  3. [Section V-B, Table II] The paper reports that TrajGEOS results are averages over five independent runs, but no standard deviations are given, and it is not stated whether the baselines were run multiple times or with what variance. Given that the R@1 margins over the best baselines are 0.86 pp (NYC), 1.02 pp (TKY), and 0.13 pp (Dallas), the state-of-the-art claim requires at least a standard deviation or a significance test on identical test instances. This is particularly important because, on NYC, the ablation TrajGEOS-woGraph (R@1 = 0.2622) falls below MTNet (0.2635), so the entire claimed advantage on NYC rests on the graph module whose comparison is affected by the preprocessing differences raised in the first comment.
minor comments (7)
  1. [Section III, Definition 3.3] The edge set is written 'E = {ei->j, ei->j, ...}' with the same ordered pair repeated; this should be a list of distinct transitions.
  2. [Section IV-D-1, Eq. (13)] In the softmax denominator, 'Pk l=1 exp(beta_j)' should be 'Pk l=1 exp(beta_l)', since the summation index is l.
  3. [Section IV-B] The phrase 'the k-th dim in the f lowof ei->j' has a spacing error, and 'an experiential value used in dropout layers' should read 'an empirical value'.
  4. [Section IV-B, Eq. (8)] The subscript in 'h^{k-1}_{i-1,s}' appears to be a typo for 'h^{k-1}_{i,s}', given that the preceding term is 'h^{k}_{i,s}'.
  5. [Table II] The TKY column header 'Recall10' is inconsistent with 'R@10' used for the other columns in the same table.
  6. [Appendix C-F] The text says 'we present our findings in Table A', but the referenced table is labeled Table VIII.
  7. [Table I] The meaning of the 'Processed' row is unclear, since no model, including TrajGEOS, uses exactly those counts (e.g., NYC 'Processed' has 1,083 users and 139,183 records, while TrajGEOS reports 1,065 users and 131,874 records).

Circularity Check

1 steps flagged · score 2.0 of 10

No derivation-level circularity; the only circular element is mild test-set hyperparameter selection on the NYC Recall@1 that is later reported as the headline NYC result.

  1. fitted input called prediction [Appendix C, Tables IV-VIII; Section V-B, Table II]
    "Although the difference in results that have different α is not large, we choose α = 0.7 as our optimal parameter based on the evaluation metric Recall@1 for the next location prediction task."

    The α=0.7 row of Table IV reports NYC location R@1 = 0.2721, exactly the Table II value for TrajGEOS; Tables V, VII, and VIII repeat this selection by Recall@1 for recent-window length and graph depths, also converging on 0.2721. So the reported NYC number is the coordinate-wise optimum of a grid search whose objective is the same Recall@1 metric that is then presented as the result. The advertised prediction is the selection criterion itself, not an independent estimate; this is mild test-set selection bias rather than a full equation-level circularity, and it mainly affects the NYC comparison.

full rationale

The paper contains no analytic derivation that could reduce a claimed result to its inputs: the model is trained by cross-entropy on sub-trajectories and evaluated on held-out chronological test sub-trajectories (Section III, Eqs. 17-18). The architecture components (graph learning, GRU short-term module, orientation mid-term module, long-term graph readout) are ablated against the same held-out labels in Table II, which gives their contributions independent empirical content. There are no load-bearing self-citations: reference [37] is mentioned only as related work and is not used to justify the method. The main legitimate concern is Appendix C, where α, recent-trajectory length, and both graph depths are selected by Recall@1 on the NYC test partition, making the reported NYC R@1 of 27.2% a selected value rather than a fresh estimate; this is a validity issue for the NYC SOTA comparison, not derivation-level circularity. The separate per-model preprocessing in Table I is a comparability concern, not a circularity concern. Overall, the central claim retains independent empirical content, so the circularity score is low.

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

This is an empirical machine learning paper, so the ledger lists the modeling choices and evaluation premises the reported numbers depend on rather than mathematical axioms. The most load-bearing items are the comparability of the per-model datasets and the test-set-based selection of alpha, recent-window length, and graph depths. The millions of gradient-trained weights in the GRU, MLPs, and GNNs are standard learned parameters and are not enumerated individually. No invented physical or conceptual entities are introduced.

free parameters (5)
  • multi-task loss weight alpha = 0.7
    Set to maximize location Recall@1 on the NYC test set (Table IV); the headline NYC R@1 depends on this choice.
  • recent trajectory length kappa = 2 weeks
    Chosen from {1,2,4,6,8} weeks using NYC test Recall@1 (Table V); directly defines the input to the orientation module.
  • EGraphSAGE layer count = 2
    Chosen from {2,3,4,5} using NYC test Recall@1 (Table VII); sets the depth of the global location embedding.
  • GraphSAGE layer count on user subgraphs = 2
    Chosen from {2,3,4,5} using NYC test Recall@1 (Table VIII); sets the depth of the user long-term preference.
  • edge dropout ratio epsilon = 0.5
    Set as an experiential value to reduce oversmoothing in the global trajectory graph (Section IV-B), not validated on a held-out split.
assumptions (5)
  • domain assumption Weekly sub-trajectory segmentation preserves the temporal structure needed for next-visit prediction.
    The entire long/mid/short preference decomposition is built on splitting records into weeks (Section III); if weekly periodicity is not the right scale, the preference signals lose meaning.
  • domain assumption The global trajectory graph built only from training transitions provides useful relational signal without leakage.
    Edge attributes (distance, transition count, 24-hour flow) are computed from training data (Section IV-B); the model must generalize to unseen test transitions, and no leakage analysis is provided.
  • ad hoc to paper The past two-to-three-week window with kappa equal to 2 captures mid-term user preference.
    Section IV-D fixes kappa to 2, and Table V shows the chosen value is tied to the NYC test set, with nearby window lengths performing comparably.
  • domain assumption Preprocessed datasets for different baselines are comparable despite different user, location, and record counts.
    Section V-A asserts the differences are minimal, but Table I shows non-identical subsets; the state-of-the-art comparison relies on this assumption.
  • domain assumption EGraphSAGE and GraphSAGE message passing with dropout learns embeddings that help downstream prediction.
    Standard GNN effectiveness is assumed for the global and user-subgraph convolutions; the ablation partially supports this by showing worse results without the graph module.

how reviews work

0 comments
Cite this review

Pith. "Pith review of TrajGEOS: Trajectory Graph Enhanced Orientation-based Sequential Network for Mobility Prediction." pith.science (2026). https://pith.science/paper/R26QBXPW

@misc{pith2026241219092,
  author       = {Pith},
  title        = {Pith review of: TrajGEOS: Trajectory Graph Enhanced Orientation-based Sequential Network for Mobility Prediction},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/R26QBXPW}},
  note         = {Machine review of arXiv:2412.19092}
}
read the original abstract

Human mobility studies how people move to access their needed resources and plays a significant role in urban planning and location-based services. As a paramount task of human mobility modeling, next location prediction is challenging because of the diversity of users' historical trajectories that gives rise to complex mobility patterns and various contexts. Deep sequential models have been widely used to predict the next location by leveraging the inherent sequentiality of trajectory data. However, they do not fully leverage the relationship between locations and fail to capture users' multi-level preferences. This work constructs a trajectory graph from users' historical traces and proposes a \textbf{Traj}ectory \textbf{G}raph \textbf{E}nhanced \textbf{O}rientation-based \textbf{S}equential network (TrajGEOS) for next-location prediction tasks. TrajGEOS introduces hierarchical graph convolution to capture location and user embeddings. Such embeddings consider not only the contextual feature of locations but also the relation between them, and serve as additional features in downstream modules. In addition, we design an orientation-based module to learn users' mid-term preferences from sequential modeling modules and their recent trajectories. Extensive experiments on three real-world LBSN datasets corroborate the value of graph and orientation-based modules and demonstrate that TrajGEOS outperforms the state-of-the-art methods on the next location prediction task.

Figures

Figures reproduced from arXiv: 2412.19092 by the authors.

Figure 1
Figure 1. (a) Implicit relationship between locations. (b) Sample users’ trajectory [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The architecture of TrajGEOS. It contains three modules: graph learning module, trajectory embedding module, and prediction module. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Hierarchical graph convolution in the graph modeling module. [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: (a) The cumulative distribution function of distance ˆ [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: (a) Relationship of historical location complexity and predicted [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: (a) The average performance of different super-categories. (b) The [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

47 extracted references · 34 canonical work pages

  1. [1]

    Collective benefits in traffic during mega events via the use of information technologies,

    Y . Xu and M. C. Gonz ´alez, “Collective benefits in traffic during mega events via the use of information technologies,” Journal of The Royal Society Interface, vol. 14, no. 129, p. 20161041, 2017

    work page 2017

  2. [2]

    Understanding congested travel in urban areas,

    S. C ¸ olak, A. Lima, and M. C. Gonz´alez, “Understanding congested travel in urban areas,” Nature communications, vol. 7, no. 1, p. 10793, 2016

    work page 2016

  3. [3]

    Planning for electric vehicle needs by coupling charging profiles with urban mobility,

    Y . Xu, S. C ¸ olak, E. C. Kara, S. J. Moura, and M. C. Gonz´alez, “Planning for electric vehicle needs by coupling charging profiles with urban mobility,” Nature Energy, vol. 3, no. 6, pp. 484–493, 2018

    2018

  4. [4]

    Planning for sustainable cities by estimating building occu- pancy with mobile phones,

    E. Barbour, C. C. Davila, S. Gupta, C. Reinhart, J. Kaur, and M. C. Gonz´alez, “Planning for sustainable cities by estimating building occu- pancy with mobile phones,” Nature communications, vol. 10, no. 1, p. 3736, 2019

    work page 2019

  5. [5]

    Emergence of urban growth patterns from human mobility behavior,

    F. Xu, Y . Li, D. Jin, J. Lu, and C. Song, “Emergence of urban growth patterns from human mobility behavior,” Nature Computational Science, vol. 1, no. 12, pp. 791–800, 2021

    work page 2021

  6. [6]

    Mobile phone location data for disasters: A review from natural hazards and epidemics,

    T. Yabe, N. K. Jones, P. S. C. Rao, M. C. Gonzalez, and S. V . Ukkusuri, “Mobile phone location data for disasters: A review from natural hazards and epidemics,” Computers, Environment and Urban Systems , vol. 94, p. 101777, 2022

    work page 2022

  7. [7]

    Limits of predictability in human mobility,

    C. Song, Z. Qu, N. Blumm, and A.-L. Barab ´asi, “Limits of predictability in human mobility,” Science, vol. 327, no. 5968, pp. 1018–1021, 2010

    work page 2010

  8. [8]

    Understanding vehicular routing behavior with location-based service data,

    Y . Xu, R. D. Clemente, and M. C. Gonz ´alez, “Understanding vehicular routing behavior with location-based service data,” EPJ Data Science , vol. 10, no. 1, pp. 1–17, 2021

    2021

Show all 47 references
  1. [9]

    Modeling user activity preference by leveraging user spatial temporal characteristics in lbsns,

    D. Yang, D. Zhang, V . W. Zheng, and Z. Yu, “Modeling user activity preference by leveraging user spatial temporal characteristics in lbsns,” IEEE Transactions on Systems, Man, and Cybernetics: Systems , vol. 45, no. 1, pp. 129–142, 2014

    2014

  2. [10]

    Hgarn: Hierarchical graph atten- tion recurrent network for human mobility prediction,

    Y . Tang, J. He, and Z. Zhao, “Hgarn: Hierarchical graph atten- tion recurrent network for human mobility prediction,” arXiv preprint arXiv:2210.07765, 2022

    2022 arXiv

  3. [11]

    Where to go next: A spatio-temporal gated network for next poi recommendation,

    P. Zhao, A. Luo, Y . Liu, J. Xu, Z. Li, F. Zhuang, V . S. Sheng, and X. Zhou, “Where to go next: A spatio-temporal gated network for next poi recommendation,” IEEE Transactions on Knowledge and Data Engineering, vol. 34, no. 5, pp. 2512–2524, 2020

    2020

  4. [12]

    Predicting the next location: A recurrent model with spatial and temporal contexts,

    Q. Liu, S. Wu, L. Wang, and T. Tan, “Predicting the next location: A recurrent model with spatial and temporal contexts,” in Thirtieth AAAI conference on artificial intelligence , 2016

    2016

  5. [13]

    Location prediction over sparse user mobility traces using rnns,

    D. Yang, B. Fankhauser, P. Rosso, and P. Cudre-Mauroux, “Location prediction over sparse user mobility traces using rnns,” in Proceedings of the Twenty-Ninth International Joint Conference on Artificial Intelli- gence, 2020, pp. 2184–2190

    2020

  6. [14]

    Hst-lstm: A hierarchical spatial-temporal long- short term memory network for location prediction

    D. Kong and F. Wu, “Hst-lstm: A hierarchical spatial-temporal long- short term memory network for location prediction.” in IJCAI, vol. 18, no. 7, 2018, pp. 2341–2347

    2018

  7. [15]

    A contextual attention recurrent architecture for context-aware venue recommendation,

    J. Manotumruksa, C. Macdonald, and I. Ounis, “A contextual attention recurrent architecture for context-aware venue recommendation,” in The 41st international ACM SIGIR conference on research & development in information retrieval , 2018, pp. 555–564

    2018

  8. [16]

    Context-aware deep model for joint mobility and time prediction,

    Y . Chen, C. Long, G. Cong, and C. Li, “Context-aware deep model for joint mobility and time prediction,” in Proceedings of the 13th International Conference on Web Search and Data Mining , 2020, pp. 106–114

    2020

  9. [17]

    Deep- move: Predicting human mobility with attentional recurrent networks,

    J. Feng, Y . Li, C. Zhang, F. Sun, F. Meng, A. Guo, and D. Jin, “Deep- move: Predicting human mobility with attentional recurrent networks,” in Proceedings of the 2018 world wide web conference, 2018, pp. 1459– 1468

    2018

  10. [18]

    Stan: Spatio-temporal attention network for next location recommendation,

    Y . Luo, Q. Liu, and Z. Liu, “Stan: Spatio-temporal attention network for next location recommendation,” in Proceedings of the Web Conference 2021, 2021, pp. 2177–2185

    2021

  11. [19]

    Getnext: trajectory flow map enhanced transformer for next poi recommendation,

    S. Yang, J. Liu, and K. Zhao, “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, 2022, pp. 1144–1153

    2022

  12. [20]

    How do you go where? improving next location prediction by learning travel mode information using transformers,

    Y . Hong, H. Martin, and M. Raubal, “How do you go where? improving next location prediction by learning travel mode information using transformers,” in Proceedings of the 30th International Conference on Advances in Geographic Information Systems , 2022, pp. 1–10

    2022

  13. [21]

    Mobtcast: Leveraging auxiliary trajectory forecasting for human mobility prediction,

    H. Xue, F. Salim, Y . Ren, and N. Oliver, “Mobtcast: Leveraging auxiliary trajectory forecasting for human mobility prediction,” Advances in Neural Information Processing Systems , vol. 34, pp. 30 380–30 391, 2021

    2021

  14. [22]

    Factorizing personalized markov chains for next-basket recommendation,

    S. Rendle, C. Freudenthaler, and L. Schmidt-Thieme, “Factorizing personalized markov chains for next-basket recommendation,” in Pro- ceedings of the 19th international conference on World wide web , 2010, pp. 811–820

    2010

  15. [23]

    Where you like to go next: Successive point-of-interest recommendation,

    C. Cheng, H. Yang, M. R. Lyu, and I. King, “Where you like to go next: Successive point-of-interest recommendation,” in Twenty-Third international joint conference on Artificial Intelligence , 2013

    2013

  16. [24]

    Exploiting geographical influence for collaborative point-of-interest recommendation,

    M. Ye, P. Yin, W.-C. Lee, and D.-L. Lee, “Exploiting geographical influence for collaborative point-of-interest recommendation,” in Pro- ceedings of the 34th international ACM SIGIR conference on Research and development in Information Retrieval , 2011, pp. 325–334

    2011

  17. [25]

    Fused matrix factorization with geographical and social influence in location-based social net- works,

    C. Cheng, H. Yang, I. King, and M. Lyu, “Fused matrix factorization with geographical and social influence in location-based social net- works,” in Proceedings of the AAAI conference on artificial intelligence , vol. 26, no. 1, 2012, pp. 17–23

    2012

  18. [26]

    Rank- geofm: A ranking based geographical factorization method for point of interest recommendation,

    X. Li, G. Cong, X.-L. Li, T.-A. N. Pham, and S. Krishnaswamy, “Rank- geofm: A ranking based geographical factorization method for point of interest recommendation,” in Proceedings of the 38th international ACM SIGIR conference on research and development in information retriev...

    2015

  19. [27]

    Geomf: joint geographical modeling and matrix factorization for point-of-interest rec- ommendation,

    D. Lian, C. Zhao, X. Xie, G. Sun, E. Chen, and Y . Rui, “Geomf: joint geographical modeling and matrix factorization for point-of-interest rec- ommendation,” in Proceedings of the 20th ACM SIGKDD international conference on Knowledge discovery and data mining , 2014, pp. 831– 840. 12

    2014

  20. [28]

    Person- alized ranking metric embedding for next new poi recommendation,

    S. Feng, X. Li, Y . Zeng, G. Cong, Y . M. Chee, and Q. Yuan, “Person- alized ranking metric embedding for next new poi recommendation,” in Twenty-Fourth International Joint Conference on Artificial Intelligence , 2015

    2015

  21. [29]

    Hme: A hyperbolic metric embedding approach for next-poi recommendation,

    S. Feng, L. V . Tran, G. Cong, L. Chen, J. Li, and F. Li, “Hme: A hyperbolic metric embedding approach for next-poi recommendation,” in Proceedings of the 43rd International ACM SIGIR Conference on Research and Development in Information Retrieval , 2020, pp. 1429– 1438

    2020

  22. [30]

    Stellar: Spatial- temporal latent ranking for successive point-of-interest recommenda- tion,

    S. Zhao, T. Zhao, H. Yang, M. R. Lyu, and I. King, “Stellar: Spatial- temporal latent ranking for successive point-of-interest recommenda- tion,” in Thirtieth AAAI conference on artificial intelligence , 2016

    2016

  23. [31]

    Long-and short-term preference learning for next poi recommendation,

    Y . Wu, K. Li, G. Zhao, and X. Qian, “Long-and short-term preference learning for next poi recommendation,” in Proceedings of the 28th ACM international conference on information and knowledge management , 2019, pp. 2301–2304

    2019

  24. [32]

    Geography-aware sequential location recommendation,

    D. Lian, Y . Wu, Y . Ge, X. Xie, and E. Chen, “Geography-aware sequential location recommendation,” in Proceedings of the 26th ACM SIGKDD international conference on knowledge discovery & data mining, 2020, pp. 2009–2019

    2020

  25. [33]

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

    K. Sun, T. Qian, T. Chen, Y . Liang, Q. V . H. Nguyen, and H. Yin, “Where to go next: Modeling long-and short-term user preferences for point-of- interest recommendation,” in Proceedings of the AAAI Conference on Artificial Intelligence, vol. 34, no. 01, 2020, pp. 214–221

    2020

  26. [34]

    Personalized long-and short-term preference learning for next poi recommendation,

    Y . Wu, K. Li, G. Zhao, and Q. Xueming, “Personalized long-and short-term preference learning for next poi recommendation,” IEEE Transactions on Knowledge and Data Engineering , 2020

    2020

  27. [35]

    Dis- entangling geographical effect for point-of-interest recommendation,

    Y . Qin, C. Gao, Y . Wang, S. Wei, D. Jin, J. Yuan, and L. Zhang, “Dis- entangling geographical effect for point-of-interest recommendation,” IEEE Transactions on Knowledge and Data Engineering , vol. 35, no. 8, pp. 7883–7897, 2022

    2022

  28. [36]

    Pre- dicting human mobility via self-supervised disentanglement learning,

    Q. Gao, J. Hong, X. Xu, P. Kuang, F. Zhou, and G. Trajcevski, “Pre- dicting human mobility via self-supervised disentanglement learning,” IEEE Transactions on Knowledge and Data Engineering , 2023

    2023

  29. [37]

    Human mobility prediction with causal and spatial-constrained multi-task network,

    Z. Huang, S. Xu, M. Wang, H. Wu, Y . Xu, and Y . Jin, “Human mobility prediction with causal and spatial-constrained multi-task network,” EPJ Data Science, vol. 13, no. 1, p. 22, 2024

    2024

  30. [38]

    Learning graph-based poi embedding for location-based recommendation,

    M. Xie, H. Yin, H. Wang, F. Xu, W. Chen, and S. Wang, “Learning graph-based poi embedding for location-based recommendation,” in Proceedings of the 25th ACM international on conference on information and knowledge management , 2016, pp. 15–24

    2016

  31. [39]

    Dynamic discovery of favorite locations in spatio-temporal social networks,

    X. Xiong, F. Xiong, J. Zhao, S. Qiao, Y . Li, and Y . Zhao, “Dynamic discovery of favorite locations in spatio-temporal social networks,” Information Processing & Management, vol. 57, no. 6, p. 102337, 2020

    2020

  32. [40]

    Attention is all you need,

    A. Vaswani, N. Shazeer, N. Parmar, J. Uszkoreit, L. Jones, A. N. Gomez, Ł. Kaiser, and I. Polosukhin, “Attention is all you need,” Advances in neural information processing systems , vol. 30, 2017

    2017

  33. [41]

    Graph-flashback network for next location recommendation,

    X. Rao, L. Chen, Y . Liu, S. Shang, B. Yao, and P. Han, “Graph-flashback network for next location recommendation,” in Proceedings of the 28th ACM SIGKDD conference on knowledge discovery and data mining , 2022, pp. 1463–1471

    2022

  34. [42]

    Spatio-temporal hypergraph learning for next poi recommendation,

    X. Yan, T. Song, Y . Jiao, J. He, J. Wang, R. Li, and W. Chu, “Spatio-temporal hypergraph learning for next poi recommendation,” in Proceedings of the 46th international ACM SIGIR conference on research and development in information retrieval , 2023, pp. 403–412

    2023

  35. [43]

    Learning time slot preferences via mobility tree for next poi recommendation,

    T. Huang, X. Pan, X. Cai, Y . Zhang, and X. Yuan, “Learning time slot preferences via mobility tree for next poi recommendation,” in Proceedings of the AAAI Conference on Artificial Intelligence , vol. 38, no. 8, 2024, pp. 8535–8543

    2024

  36. [44]

    Handling missing data with graph representation learning,

    J. You, X. Ma, Y . Ding, M. J. Kochenderfer, and J. Leskovec, “Handling missing data with graph representation learning,” Advances in Neural Information Processing Systems , vol. 33, pp. 19 075–19 087, 2020

    2020

  37. [45]

    Inductive representation learning on large graphs,

    W. Hamilton, Z. Ying, and J. Leskovec, “Inductive representation learning on large graphs,” Advances in neural information processing systems, vol. 30, 2017

    2017

  38. [46]

    Time2vec: Learning a vector representation of time,

    S. M. Kazemi, R. Goel, S. Eghbali, J. Ramanan, J. Sahota, S. Thakur, S. Wu, C. Smyth, P. Poupart, and M. Brubaker, “Time2vec: Learning a vector representation of time,” arXiv preprint arXiv:1907.05321 , 2019

    1907 arXiv

  39. [47]

    Friendship and mobility: user movement in location-based social networks,

    E. Cho, S. A. Myers, and J. 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, 2011, pp. 1082–1090

    2011

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.