Search Results (41)

The greatest challenge of location recommendation is data sparsity. Some works introduce auxiliary information such as social links and semantics to enrich the representation. However, there are still two main limitations. Firstly, existing models only investigate how a person chooses POIs (Point-of-Interests), highlighting the user interest in locations from a human perspective, limited by users’ sparse trajectory. Secondly, they recommend POIs by one’s trajectory independently. However, it is hard to uncover specific behavior patterns or daily routines individually for each user. To address these problems, we develop a novel method for the next location recommendation via bi-directional inference and similarity-aware enhancement (SimBIN). Specifically, it considers the location recommendation task from both the human and location view, conducting the next POI inference and the next visitor inference synchronously. In particular, we employ the adaptive learning strategy to refine the model based on the conformity of two inference parts. Moreover, for the POI inference, we take the highly similar users to approximate a user’s behavior for enriching the next POI candidates. It is similar for the user inference. Experiments show that the proposed SimBIN significantly outperforms the state-of-the-art approaches. Furthermore, each key component can be refined and generally used in other frameworks. By establishing a bidirectional probabilistic collaboration framework, this study not only advances the theoretical understanding of mutual regularization in spatial-temporal modeling, but also provides a possible solution for personalized location-aware services and urban trajectory planning. Full article

The widespread use of mobile Internet- and location-based services has generated large-scale check-in data in location-based social networks, creating opportunities for intelligent urban-mobility analysis and personalized mobility services. Making the next point-of-interest (POI) recommendation is an important task in this setting because it supports context-aware destination suggestion, travel assistance, and smart mobility services. However, existing methods still face challenges in jointly modeling higher-order mobility patterns, uneven time intervals, geographic reachability, and fine-grained intra-day temporal regularities. To address these issues, this paper proposes ASTHN, an Adaptive Spatio-Temporal Hypergraph Network for next POI recommendation. ASTHN constructs three fine-grained spatio-temporal context hypergraphs from minimum time interval, spatial proximity, and hourly preference, and uses hypergraph neural networks to learn view-specific POI representations. A context-adaptive fusion module then aligns and integrates multi-source spatio-temporal signals, while an ST-GRU with spatio-temporal gates captures dynamic trajectory evolution. Temperature scaling is further applied at the output layer to alleviate overly concentrated score distributions. Experiments on Foursquare-NYC and Foursquare-TKY show that ASTHN consistently outperforms representative baselines. With results reported as mean ± std over three random seeds, ASTHN improves over the strongest baseline by 3.79%, 14.62%, 2.28%, and 1.24% on NYC in Recall@5, Recall@10, NDCG@5, and NDCG@10, respectively. On TKY, the corresponding improvements are 5.83%, 37.20%, 13.86%, and 20.49%. Ablation, parameter, complexity, and application-oriented case analyses further demonstrate the effectiveness, stability, and practical usability of ASTHN for next POI recommendation in urban-mobility scenarios. Full article

Understanding and predicting how tourists move through a city is a challenging task, as it involves a complex interplay of spatial, temporal, and social factors. Traditional recommender systems often rely on structured data, trying to capture the nature of the problem. However, recent advances in Large Language Models (LLMs) open new possibilities for reasoning over richer, text-based representations of user context, even without a dedicated pre-training phase. In this study, we investigate the potential of LLMs to interpret and predict tourist movements in a real-world application scenario involving tourist visits to Verona, a municipality in Northern Italy, between 2014 and 2023. We propose an incremental prompt engineering approach that gradually enriches the model input, from spatial features alone to richer behavioral information, including visit histories, time information, and user cluster patterns. The approach is evaluated using six open-source models, enabling us to compare their accuracy and efficiency across various levels of contextual enrichment. The results provide a first insight about the abilities of LLMs to incorporate spatio-temporal contextual factors, thus improving predictions, while maintaining computational efficiency. The analysis of the model-generated explanations completes the picture by adding an interpretability dimension that most existing next-PoI prediction solutions lack. Overall, the study demonstrates the potential of LLMs to integrate multiple contextual dimensions in tourism mobility, highlighting the possibility of a more text-oriented, adaptive, and explainable T-RS. Full article

With the rapid advancement of location-based services, next Point-of-Interest (POI) recommendation has emerged as a critical task in personalized mobility modeling and recommendation systems. It aims to predict users’ future locations based on their historical trajectories, thereby enhancing the personalization and intelligence of recommendation systems. Despite the promising progress, two key challenges remain insufficiently addressed. First, many existing methods overlook the dynamic evolution of user trajectories across multiple perspectives, resulting in entangled representations that fail to capture user intent accurately. Second, they often ignore the latent synergy across diverse perspectives, which limits the effective utilization of complementary information for recommendation. To address these issues, we propose a novel framework called MRHL. MRHL constructs multiple hypergraphs to represent distinct views of user behavior, including interaction frequency, time decay, and geographical proximity. An enhanced hypergraph convolutional network is employed to effectively model the high-order relationships within them. We propose a cascaded enhancement fusion mechanism that progressively integrates multi-view hypergraph representations to enrich the semantic information of user representations. In addition, a multi-relational contrastive learning strategy is developed to capture the consistent signals across different views, thereby enhancing the robustness and discriminative capability of user and POI representations. Extensive experiments on three public datasets consistently demonstrate that MRHL outperforms a range of strong baselines. Full article

Next Point-of-Interest (POI) recommendation focuses on predicting a user’s subsequent location based on historical check-in data. In practice, however, check-in logs frequently contain uncertain records in which ambiguous spatial, temporal, or behavioral information obscures the underlying mobility regularities, thereby degrading prediction performance. To address this challenge, this study first infers user preferences from historical trajectories and reweights transition importance based on temporal and spatial proximity. It then models transition relationships using three complementary feature dimensions: POI category, spatial area, and routine versus non-routine behavioral patterns. Using transition probability analysis, feature-level dependencies in user mobility are systematically investigated. The findings demonstrate that these transition features contribute unevenly to predictive performance, with area-based transitions yielding the strongest results when used in isolation. Nonetheless, their joint integration consistently achieves the highest accuracy, underscoring the critical role of transition-aware modeling. Across two real-world datasets, the proposed framework consistently achieves state-of-the-art performance in top-ranked accuracy (Recall@1) and ranking quality (NDCG@1), while delivering competitive effectiveness at higher cutoff values ($k=3$ and $k=5$). Notably, on the NYC dataset, MTF-POI achieves the highest Recall@1 (+19.01% over the strongest baseline) with a marginal trade-off at Recall@3, reflecting the framework’s design emphasis on precise next-step prediction. Full article

Next Point-of-Interest (POI) recommendations are pivotal for enhancing location-based services; however, accurate prediction remains challenging due to the complex interplay between dynamic user preferences and spatiotemporal constraints. Existing graph-sequence hybrids often fail to unify these dimensions, typically treating temporal contexts as disjoint features or neglecting implicit collaborative signals within sparse user trajectories. This fragmentation limits the ability to capture high-order dependencies in user mobility. To address these challenges, we propose UPTRec, a unified framework that synergizes social, spatial, and temporal reasoning. UPTRec constructs a TF-IDF-weighted user similarity graph to recover latent social connections and a flow-based POI-transition graph to encode sequential mobility patterns. These structural priors are fused with fine-grained temporal-category embeddings (utilizing Time2Vec and periodic encoding) via a multi-layer Transformer encoder to comprehensively capture user behavior. Extensive experiments on three real-world datasets (NYC, TKY, and CA) demonstrate that UPTRec achieves state-of-the-art performance among the compared baselines under the same experimental settings. On the NYC dataset, UPTRec yields a Top-1 Accuracy of 25.76% and a Mean Reciprocal Rank (MRR) of 0.3879, representing a relative improvement of 5.8% and 7.1% over the strongest baseline (GETNext). These results validate the efficacy of jointly modeling collaborative and spatiotemporal dependencies. Full article

With the proliferation of mobile internet and location-based services, location-based social networks (LBSNs) have accumulated extensive user check-in data, driving the advancement of next Point-of-Interest (POI) recommendation systems. Although existing approaches can model sequential dependencies and spatio-temporal patterns, they often fail to fully capture users’ dynamic preferences under varying spatio-temporal contexts and lack effective integration of fine-grained semantic information. To address these limitations, this paper proposes Language-Guided Spatio-Temporal Context Learning for Next POI Recommendation (LSCNP). It employs a pre-trained BERT model to encode multi-dimensional spatio-temporal context—including geographic coordinates, visiting hours, and surrounding POI categories—into structured textual sequences for semantic understanding; constructs dual-graph structures to model spatial constraints and user transition patterns; and introduces a contrastive learning module to align spatio-temporal context with POI features, enhancing the discriminability of representations. A Transformer-based sequential encoder is adopted to capture long-range dependencies, while a neural matrix factorization decoder generates final recommendations. Experiments on three real-world LBSN datasets demonstrate that LSCNP consistently outperforms state-of-the-art baselines. Ablation studies and hyperparameter analyses further validate the contribution of each component to the overall performance. Full article

Existing point-of-interest (POI) recommendation methods often fail to capture complex contextual dependencies and suffer from severe data sparsity in location-based social networks (LBSNs). To address these limitations, this study proposes a Context-Aware Knowledge Graph Learning (CKGL) method that integrates multi-dimensional semantic information, spatio-temporal dependencies, and social relationships into a unified knowledge graph framework. First, the Context-Aware Knowledge Graph Construction (CKGC) module builds a unified POI knowledge graph that captures heterogeneous relationships among users, POIs, regions of interest (ROIs), and social links. Then, the Context-Aware Knowledge Graph Embedding (CKGE) module, based on the Translational Distance Model with Relation-Specific Spaces (TransR), learns relation-specific embeddings of entities to preserve heterogeneous semantics. Next, a Spatio-Temporal Gated Graph Neural Network (STG-GNN) captures temporal dynamics and spatial dependencies in user check-in behaviors, while the Relation-Aware Graph Attention Network (RA-GAT) enhances multi-relational reasoning and information aggregation across heterogeneous relations. Extensive experiments on two real-world LBSN datasets, Gowalla and Brightkite, demonstrate that CKGL significantly outperforms several baseline models on Recall and Normalized Discounted Cumulative Gain (NDCG), validating its effectiveness in capturing contextual semantics and improving recommendation accuracy under sparse and complex scenarios. Full article

The objective of the next POI recommendation is using the historical check-in sequences of users to learn the preferences and habits of users, providing a list of POIs that users will be inclined to visit next. Then, there are some limitations in existing POI recommendation algorithms. On the one hand, after obtaining the user’s preferences for the current period, if we consider the entire historical check-in sequence, including future check-in information, it is susceptible to the influence of noisy data, thereby reducing the accuracy of recommendations. On the other hand, the current methods generally rely on modeling long- and short-term preferences within a fixed time window, which possibly leads to an inability to capture users’ behavior characteristics at different time scales. As a result, we proposed a Multimodal Temporal Fusion for Next POI Recommendation(MTFNR). Firstly, to understand users’ preferences and habits at different periods, multiple hypergraph neural networks are constructed to analyze user behavior patterns at different stages, and in order to avoid introducing interference factors, only the check-in sequences visited in the current period are considered to reduce the impact of noise on the model’s recommendation performance. Secondly, modeling the next POI recommendation task through the fusion of time information and long- and short-term preferences in order to gain a more comprehensive understanding of users’ preferences and habits, enhance the timeliness of recommendations, and improve the accuracy of recommendations. Lastly, introducing spatio-temporal interval information into the GRU model, capturing dependencies in sequences to improve the overall performance of the model. Extensive experiments on the real LBSN datasets demonstrated the superior performance of the MTFNR model. The experimental results indicate that Top-10 recall improved 2.81% to 15.97% compared to current methods. Full article

Next Point-of-Interest (POI) recommendation is a crucial task in personalized location-based services, aiming to predict the next POI that a user might visit based on their historical trajectories. Although sequence models and Graph Neural Networks (GNNs) have achieved significant success, they often overlook the diversity and dynamics of user preferences. To address these issues, researchers have begun to employ Hypergraph Convolutional Networks (HGCNs) for disentangled representation learning. However, two critical problems have received less attention: (1) the limited expressive capacity of conventional hypergraph convolution layers, which restricts the modeling of complex nonlinear user–POI preference interactions and consequently weakens generalization performance, and (2) the inadequate utilization of contrastive learning mechanisms, which prevents fully capturing cross-view collaborative signals and limits the exploitation of complementary multi-view information. To tackle these challenges, we propose a Nonlinear Enhanced Disentangled Contrastive Hypergraph Learning (NE-DCHL) for next POI recommendation. The proposed model enhances nonlinear modeling capability and generalization by integrating ReLU activation, residual connections, and dropout regularization within the hypergraph convolution layer. A K-Nearest Neighbor (KNN)-based weighted adjacency matrix is employed to construct the geographical-view hypergraph, reducing computational complexity while maintaining essential spatial correlations. Moreover, a mini-batch InfoNCE loss and the GRACE (deep GRAph Contrastive rEpresentation learning) framework are utilized to improve efficiency and cross-view collaboration. Extensive experiments on two real-world datasets demonstrate that NE-DCHL consistently outperforms the original DCHL and other state-of-the-art approaches. Full article

The rapid expansion of mobile user behavior data has made next-Point-of-Interest (POI) recommendation increasingly vital for enhancing personalized location-based services. However, the non-uniform spatio-temporal distribution of user behavior poses significant challenges to recommendation performance. Most existing methods neglect this fundamental issue at the distribution level, while conventional data augmentation strategies fall short in optimizing spatio-temporal distribution properties. To tackle this problem, we propose a spatio-temporal Distribution Calibration framework for next-POI Recommendation (DCal-Rec), which optimizes behavioral sequence distributions through disentangled spatial and temporal operator pools. This is combined with a dual-constraint mechanism that incorporates both distribution and interest information to maintain semantic consistency. Furthermore, a multi-channel contrastive learning paradigm is introduced to jointly optimize the recommendation and contrastive tasks under a unified training objective, thereby improving the model’s generalization capability. Experimental results on three public real-world datasets demonstrate that DCal-Rec significantly outperforms baseline methods across various evaluation metrics. Full article

Next-point-of-interest (POI) recommendation aims to model user preferences based on historical information to predict future mobility behavior, which has significant application value in fields such as urban planning, traffic management, and optimizing business decisions. However, existing methods often overlook the differences in location, time, and category information features, fail to fully utilize information from various modalities, and lack effective solutions for addressing users’ incidental behavior. Additionally, existing methods are somewhat lacking in capturing users’ personalized preferences. To address these issues, we propose a new method called Multi-Branch Spectral Network with Contrastive Learning (MBSCL-Net) for next-POI recommendation. We use a multihead attention mechanism to separately capture the distinct features of location, time, and category information, and then fuse the captured features to effectively integrate cross-modal features, avoid feature confusion, and achieve effective modeling of multi-modal information. We propose converting the time-domain information of user check-ins into frequency-domain information through Fourier transformation, directly enhancing the low-frequency signals of users’ periodic behavior and suppressing occasional high-frequency noise, thereby greatly alleviating noise interference caused by the introduction of too much information. Additionally, we introduced contrastive learning loss to distinguish user behavior patterns and better model personalized preferences. Extensive experiments on two real-world datasets demonstrate that MBSCL-Net outperforms state-of-the-art (SOTA) methods. Full article

Next Point-of-Interest (POI) recommendation is aimed at predicting users’ future visits based on their current status and historical check-in records, providing convenience to users and potential profits to businesses. The Graph Neural Network (GNN) has become a common approach for this task due to the capabilities of modeling relations between nodes in a global perspective. However, most existing studies overlook the more prevalent heterogeneous relations in real-world scenarios, and manually constructed graphs may suffer from inaccuracies. To address these limitations, we propose a model called Heterogeneous Graph Structure Learning for Next POI Recommendation (HGSL-POI), which integrates three key components: heterogeneous graph contrastive learning, graph structure learning, and sequence modeling. The model first employs meta-path-based subgraphs and the user–POI interaction graph to obtain initial representations of users and POIs. Based on these representations, it reconstructs the subgraphs through graph structure learning. Finally, based on the embeddings from the reconstructed graphs, sequence modeling incorporating graph neural networks captures users’ sequential preferences to make recommendations. Experimental results on real-world datasets demonstrate the effectiveness of the proposed model. Additional studies confirm its robustness and superior performance across diverse recommendation tasks. Full article

Next point-of-interest (POI) recommendation aims to predict users’ future visitation intentions based on historical check-in trajectories. However, this task faces significant challenges, including coarse-grained user interest representation, insufficient social modeling, sparse check-in data, and the insufficient learning of contextual patterns. To address this, we propose a model that combines check-in trajectory information with user friendship relationships and uses a Transformer architecture for prediction (TraFriendFormer). Our approach begins with the construction of trajectory flow graphs using graph convolutional networks (GCNs) to globally capture POI correlations across both spatial and temporal dimensions. In parallel, we design an integrated social graph that combines explicit friendships with implicit interaction patterns, in which GraphSAGE aggregates neighborhood information to generate enriched user embeddings. Finally, we fuse the POI embeddings, user embeddings, timestamp embeddings, and category embeddings and input them into the Transformer architecture. Through the self-attention mechanism, the model captures the complex temporal relationships in the check-in sequence. We validate the effectiveness of TraFriendFormer on two real-world datasets (FourSquare and Gowalla). The experimental results show that TraFriendFormer achieves an average improvement of 10.3% to 37.2% in metrics such as Acc@k and MRR compared to the selected state-of-the-art baselines. Full article