Accurate travel time estimation (TTE) plays a crucial role in intelligent transportation systems. However, it remains challenging due to heterogeneous data sources and complex traffic dynamics. Moreover, traditional approaches typically convert trajectory data into fixed-length representations. This overlooks the inherent variability of real-world motion patterns, often resulting in information loss and redundancy. To address these challenges, this paper introduces the Multimodal Dynamic Trajectory Integration (MDTI) framework--a novel multimodal trajectory representation learning approach that integrates GPS sequences, grid trajectories, and road network constraints to enhance the performance of TTE. MDTI employs modality-specific encoders and a multimodal fusion module to capture complementary spatial, temporal, and topological semantics, while a dynamic trajectory modeling mechanism adaptively regulates information density for trajectories of varying lengths. Two self-supervised pretraining objectives, named contrastive alignment and masked language modeling, further strengthen multimodal consistency and contextual understanding. Extensive experiments on three real-world datasets demonstrate that MDTI consistently outperforms state-of-the-art baselines, confirming its robustness and strong generalization abilities. The code is publicly available at: https://github.com/City-Computing/MDTI.

Transformer structures have been widely used in sequential recommender systems (SRS). However, as user interaction histories increase, computational time and memory requirements also grow. This is mainly caused by the standard attention mechanism. Although there exist many methods employing efficient attention and SSM-based models, these approaches struggle to effectively model long sequences and may exhibit unstable performance on short sequences. To address these challenges, we design a sparse attention mechanism, BlossomRec, which models both long-term and short-term user interests through attention computation to achieve stable performance across sequences of varying lengths. Specifically, we categorize user interests in recommendation systems into long-term and short-term interests, and compute them using two distinct sparse attention patterns, with the results combined through a learnable gated output. Theoretically, it significantly reduces the number of interactions participating in attention computation. Extensive experiments on four public datasets demonstrate that BlossomRec, when integrated with state-of-the-art Transformer-based models, achieves comparable or even superior performance while significantly reducing memory usage, providing strong evidence of BlossomRec's efficiency and effectiveness. The code is available at https://github.com/Applied-Machine-Learning-Lab/WWW2026_BlossomRec.

As urban residents demand higher travel quality, vehicle dispatch has become a critical component of online ride-hailing services. However, current vehicle dispatch systems struggle to navigate the complexities of urban traffic dynamics, including unpredictable traffic conditions, diverse driver behaviors, and fluctuating supply and demand patterns. These challenges have resulted in travel difficulties for passengers in certain areas, while many drivers in other areas are unable to secure orders, leading to a decline in the overall quality of urban transportation services. To address these issues, this paper introduces GARLIC: a framework of GPT-Augmented Reinforcement Learning with Intelligent Control for vehicle dispatching. GARLIC utilizes multiview graphs to capture hierarchical traffic states, and learns a dynamic reward function that accounts for individual driving behaviors. The framework further integrates a GPT model trained with a custom loss function to enable high-precision predictions and optimize dispatching policies in real-world scenarios. Experiments conducted on two real-world datasets demonstrate that GARLIC effectively aligns with driver behaviors while reducing the empty load rate of vehicles.

Federated Graph Learning (FGL) has emerged as a key paradigm for distributed graph machine learning, enabling cross-domain graph collaborative modeling while preserving data privacy. However, existing methods face two major bottlenecks: the structural heterogeneity discrepancy of graph data among clients weakens the generalization ability of the global model; and model heterogeneity leads to inefficient knowledge sharing and complex global aggregation. To address these issues, we propose FedGVD, an efficient framework that constructs a global perspective through data condensation and server-side virtual node generation, which not only preserves the semantic equivalence of the original data but also avoids privacy leakage. Subsequently, by distributing low-dimensional generalizable knowledge for unidirectional distillation, FedGVD enables local models to absorb global knowledge without transmitting local parameters, thus breaking through the challenges of data and structural heterogeneity as well as model heterogeneity. This innovative approach ensures privacy-preserving and efficient federated graph collaboration. Experiments show that FedGVD maintains excellent performance in heterogeneous model scenarios while significantly improving communication efficiency, offering a new approach for privacy-preserving collaborative modeling in FGL. The code is available at https://github.com/Jasonxx4/FedGVD.

Large Language Models (LLMs) possess remarkable generalization capabilities but struggle with multi-task adaptation, particularly in balancing knowledge retention with task-specific specialization. Conventional fine-tuning methods suffer from catastrophic forgetting and substantial resource consumption, while existing parameter-efficient methods perform suboptimally in complex multi-task scenarios. To address this, we propose Contextual Attention Modulation (CAM), a novel mechanism that dynamically modulates the representations of self-attention modules in LLMs. CAM enhances task-specific features while preserving general knowledge, thereby facilitating more effective and efficient adaptation. For effective multi-task adaptation, CAM is integrated into our Hybrid Contextual Attention Modulation (HyCAM) framework, which combines a shared, full-parameter CAM module with multiple specialized, lightweight CAM modules, enhanced by a dynamic routing strategy for adaptive knowledge fusion. Extensive experiments on heterogeneous tasks, including question answering, code generation, and logical reasoning, demonstrate that our approach significantly outperforms existing approaches, achieving an average performance improvement of 3.65%. The implemented code and data are available to ease reproducibility at https://github.com/Applied-Machine-Learning-Lab/HyCAM.

Spatio-temporal graph modeling is widely applied to spatio-temporal data, analyzing the relationships between data to achieve accurate predictions. However, despite the excellent predictive performance of increasingly complex models, their intricate architectures result in significant memory overhead and computational complexity when handling spatio-temporal data, which limits their practical applications. To address these challenges, we propose a plug-and-play SubGraph Learning (SGL) method to reduce the memory overhead without compromising performance. Specifically, we introduce a SubGraph Partition Module (SGPM), which leverages a set of learnable memory vectors to select node groups with similar features from the graph, effectively partitioning the graph into smaller subgraphs. Noting that partitioning the graph may lead to feature redundancy, as overlapping information across subgraphs can occur. To overcome this, we design a SubGraph Feature Aggregation Module (SGFAM), which mitigates redundancy by averaging node features from different subgraphs. Experiments on four traffic network datasets of various scales demonstrate that SGL can significantly reduce memory overhead, achieving up to a 56.4\% reduction in average GPU memory overhead, while maintaining robust prediction performance. The source code is available at https://github.com/wengwenchao123/SubGraph-Learning.

Visual geo-localization demands in-depth knowledge and advanced reasoning skills to associate images with precise real-world geo-graphic locations. Existing image database retrieval methods are limited by the impracticality of storing sufficient visual records of global landmarks. Recently, Large Vision-Language Models (LVLMs) have demonstrated the capability of geo-localization through Visual Question Answering (VQA), enabling a solution that does not require external geo-tagged image records. However, the performance of a single LVLM is still limited by its intrinsic knowledge and reasoning capabilities. To address these challenges, we introduce smileGeo, a novel visual geo-localization framework that leverages multiple Internet-enabled LVLM agents operating within an agent-based architecture. By facilitating inter-agent communication, smileGeo integrates the inherent knowledge of these agents with additional retrieved information, enhancing the ability to effectively localize images. Furthermore, our framework incorporates a dynamic learning strategy that optimizes agent communication, reducing redundant interactions and enhancing overall system efficiency. To validate the effectiveness of the proposed framework, we conducted experiments on three different datasets, and the results show that our approach significantly outperforms current state-of-the-art methods. The source code is available at https://github.com/Applied-Machine-Learning-Lab/smileGeo.

Multi-modal recommender systems have become indispensable in modern applications. Despite promising results, existing methods have two key limitations. First, in item characteristic modeling, they rely solely on the merchant’s description and often neglect customer reviews, leading to biased item quality assessment. Second, in user preference modeling, they focus mainly on one-hop user-item interactions in the interaction graph, overlooking multi-hop interactions, which limits the understanding of user preferences. To address these issues, we propose a Joint Content and Interaction Augmented Framework (JCIAF) for multi-modal recommendation. Specifically, we leverage large language models to extract valuable insights from user reviews, integrating this with the merchant’s description to form a more comprehensive textual representation of the item. This enriched description provides a balanced foundation for item characteristic modeling. Next, we enhance the user-item interaction graph with two additional interaction types: user-item-item (two-hop) and user-item-user-item (three-hop), which offer augmented views for more thorough user preference modeling. We apply a diffusion-based method across the three augmented graphs and introduce an online knowledge distillation mechanism to enable cross-graph learning. Extensive experiments on three real-world datasets demonstrate the effectiveness of our proposed method. The source code is accessible at https://github.com/jjlinnn/JClAF.git.

Recent advancements in Large Language Models (LLMs) have emphasized the critical role of fine-tuning (FT) techniques in adapting LLMs to specific tasks, especially when retraining from scratch is computationally infeasible. Fine-tuning enables LLMs to leverage task- or domain-specific data, producing models that more effectively meet the requirements of targeted applications. However, conventional FT approaches often suffer from catastrophic forgetting and suboptimal data efficiency, limiting their real-world applicability. To address these challenges, this paper proposes DEAL, a novel framework that integrates Low-Rank Adaptation (LoRA) with a continuous fine-tuning strategy. By incorporating knowledge retention and adaptive parameter update modules, the framework mitigates the limitations of existing FT methods while maintaining efficiency. Experiments on 15 diverse datasets show that DEAL consistently outperforms baseline methods, yielding substantial gains in task accuracy and resource efficiency. These findings demonstrate the potential of our approach to advance continual adaptation in LLMs by enhancing task performance while improving resource efficiency. The source code is publicly available at https://github.com/Applied-Machine-Learning-Lab/DEAL.

Spatio-temporal data are foundational to understanding and modeling dynamic real-world phenomena such as human mobility, traffic flow, epidemic spread, and urban dynamics.With the growing availability of location-aware web data and the rise of intelligent urban infrastructures, analyzing spatio-temporal patterns has become both highly valuable and technically challenging.This tutorial provides a comprehensive overview of spatio-temporal data analytics, unifying perspectives from data management, research methodology, and emerging foundation models.We begin with a review of spatio-temporal data management systems, introducing the core data models, spatial-temporal indexing techniques, and scalable architectures for storing and querying large-scale mobility data.We then delve into trajectory learning, covering methods for prediction, generation, and reconstruction of movement sequences at the individual level.Next, we explore spatio-temporal graph learning, which focuses on forecasting region-level dynamics using dynamic graph neural networks.Multi-region, multi-task, and multi-domain spatio-temporal learning will be identified and introduced in detail.Finally, we present advanced learning frameworks that integrate federated learning, continual learning, and LLM-based approaches to build privacy-preserving, scalable, and adaptive spatio-temporal models.Through the lens of recent methodological and systemlevel advances, this tutorial bridges algorithmic design and practical deployment of spatio-temporal learning systems.It is suitable for researchers and practitioners working in machine learning, data mining, geospatial analysis, and intelligent systems.

Job recommender systems are crucial for aligning job opportunities with job-seekers in online job-seeking. However, users tend to adjust their job preferences to secure employment opportunities continually, which limits the performance of job recommendations. The inherent frequency of preference drift poses a challenge to promptly and precisely capture user preferences. To address this issue, we propose a novel session-based framework, BISTRO, to timely model user preference through fusion learning of semantic and behavioral information. Specifically, BISTRO is composed of three stages: 1) coarse-grained semantic clustering, 2) fine-grained job preference extraction, and 3) personalized top-k job recommendation. Initially, BISTRO segments the user interaction sequence into sessions and leverages session-based semantic clustering to achieve broad identification of person-job matching. Subsequently, we design a hypergraph wavelet learning method to capture the nuanced job preference drift. To mitigate the effect of noise in interactions caused by frequent preference drift, we innovatively propose an adaptive wavelet filtering technique to remove noisy interaction. Finally, a recurrent neural network is utilized to analyze session-based interaction for inferring personalized preferences. Extensive experiments on three real-world offline recruitment datasets demonstrate the significant performances of our framework. Significantly, BISTRO also excels in online experiments, affirming its effectiveness in live recruitment settings. This dual success underscores the robustness and adaptability of BISTRO. The source code is available at https://github.com/Applied-Machine-Learning-Lab/BISTRO.

Worldwide geolocalization aims to locate the precise location at the coordinate level of photos taken anywhere on the Earth. It is very challenging due to 1) the difficulty of capturing subtle location-aware visual semantics, and 2) the heterogeneous geographical distribution of image data. As a result, existing studies have clear limitations when scaled to a worldwide context. They may easily confuse distant images with similar visual contents, or cannot adapt to various locations worldwide with different amounts of relevant data. To resolve these limitations, we propose **G3**, a novel framework based on Retrieval-Augmented Generation (RAG). In particular, G3 consists of three steps, i.e., **G**eo-alignment, **G**eo-diversification, and **G**eo-verification to optimize both retrieval and generation phases of worldwide geolocalization. During Geo-alignment, our solution jointly learns expressive multi-modal representations for images, GPS and textual descriptions, which allows us to capture location-aware semantics for retrieving nearby images for a given query. During Geo-diversification, we leverage a prompt ensembling method that is robust to inconsistent retrieval performance for different image queries. Finally, we combine both retrieved and generated GPS candidates in Geo-verification for location prediction. Experiments on two well-established datasets IM2GPS3k and YFCC4k verify the superiority of G3 compared to other state-of-the-art methods. Our code is available online [https://github.com/Applied-Machine-Learning-Lab/G3](https://github.com/Applied-Machine-Learning-Lab/G3) for reproduction.

Traffic signal control plays a pivotal role in the management of urban traffic flow. With the rapid advancement of reinforcement learning, the development of signal control methods has seen a significant boost. However, a major challenge in implementing these methods is ensuring that signal lights do not change abruptly, as this can lead to traffic accidents. To mitigate this risk, a time-delay is introduced in the implementation of control actions, but usually has a negative impact on the overall efficacy of the control policy. To address this challenge, this paper presents a novel Traffic Signal Control Framework (PRLight), which leverages an On-policy Traffic Control Model (OTCM) and an Online Traffic Prediction Model (OTPM) to achieve efficient and real-time control of traffic signals. The framework collects multi-source traffic information from a local-view graph in real-time and employs a novel fast attention mechanism to extract relevant traffic features. To be specific, OTCM utilizes the predicted traffic state as input, eliminating the need for communication with other agents and maximizing computational efficiency while ensuring that the most relevant information is used for signal control. The proposed framework was evaluated on both simulated and real-world road networks and compared to various state-of-the-art methods, demonstrating its effectiveness in preventing traffic congestion and accidents.

The recommendation of taxi pickup locations plays an important role for drivers in carrying passengers efficiently. In addition, the emergence of the Internet of Vehicles provides technical support for it. However, existing recommendation methods do not model dynamic global positioning system information well and in real-time. In this article, we propose a spatio-temporal digraph convolutional network (STDCN) model. First, the pickup and drop-off locations are modeled into a directed spatio-temporal graph as input to the model. The correlation between each node is calculated as a unified edge weight based on the gray relational analysis. Then, the STDCN is used for dynamic spatio-temporal feature extraction. Finally, the edge-cloud collaboration framework is adopted to recommend local taxi pickup locations in real-time. The experimental results show that the proposed method is better than competing methods in terms of effectiveness and efficiency, and it shows good industrial conversion application prospects.

Congestion recognition is necessary for vehicle routing, traffic control, and many other applications in intelligent transportation systems. Besides, traffic facilities in the three-dimensional road network, which contains the fundamental spatiotemporal features for congestion recognition, provides multi-source traffic information. To exploit these traffic big data, in this paper, we propose an attention mechanism-based digraph convolution network (ADGCN) enabled framework to tackle the congestion recognition problem. It can be divided into two parts, spatial relevance modeling and temporal relevance modeling. At first, the representation incorporates spatiotemporal traffic information with the three-dimensional urban network, and partially decouples the global network topology to a single-knot digraph. Then a digraph-based convolution network is used to capture high-order spatial features. Finally, to proceed with time-series features, the multi-modal attention mechanism is introduced to catch the long-range temporal dependence and the congestion classifier is defined accordingly. This distinguishes the proposed model from the conventional congestion recognition methods. Comprehensive experiments are conducted based on real traffic data. The results demonstrate the advantages of the proposed framework over the existing spatiotemporal analysis methods.