Urban computing harnesses big data to decode complex urban dynamics and revolutionize location-based services. Traditional approaches have treated geospatial prediction tasks (e.g., estimating socio-economic indicators) and retrieval tasks (e.g., querying geographic objects) as isolated challenges, necessitating separate models with distinct training objectives. This fragmentation imposes significant computational burdens and limits cross-task synergy, despite advances in representation learning and multi-task foundation models. We present UrbanSparse, a pioneering framework that unifies geospatial prediction and retrieval through a novel sparse-dense representation architecture. By synergistically combining these tasks, UrbanSparse eliminates redundant systems while amplifying their mutual strengths. Our approach introduces two innovations: (1) Bloom filter-based sparse encodings that compress high-sparsity geographic queries and fine-grained text terms for retrieval effectiveness, and (2) a dense semantic codebook that captures granular urban features to boost prediction accuracy. A two-view contrastive learning mechanism further bridges urban objects, regions, and contexts. Experiments on real-world datasets demonstrate 25.16% gains in prediction accuracy and 20.76% improvements in retrieval precision over state-of-the-art baselines, alongside 65.97% faster training. These advantages position UrbanSparse as a scalable solution for large urban datasets. To our knowledge, this is the first unified framework bridging geospatial prediction and retrieval, opening new frontiers in data-driven urban intelligence.

Geographic Information Retrieval (GIR) systems process text queries with geographic location to identify relevant geographic objects for users. Although recent advancements have leveraged Pre-trained Language Models (PLMs) for their robust semantic comprehension, these models typically depend on extensive labeled queries and require considerable computational resources. Deviating from this prevailing trend, we propose GeoBloom, a lightweight framework that surpasses the effectiveness of PLMs with fewer or no labeled queries, with remarkable efficiency in both time and space. GeoBloom tackles critical challenges such as the lack of labeled queries, low data (labeled) efficiency, and high computational demands. At its core, it employs Bloom filters to encode text at a fine-grained term level and uses intersecting bits to create a robust unsupervised text similarity metric. A specialized Bloom Filter Evaluator is proposed to assess the importance of each intersecting bit, focusing on those associated with ground truth, improving effectiveness with fewer training labels. For enhanced search efficiency, the evaluator exploits the inherent sparsity of Bloom filters, achieving remarkably low time and space complexities. This efficiency is further boosted by a tree-based index that partitions the search space while preserving effectiveness. Extensive experiments show that GeoBloom surpasses state-of-the-art baselines in both unsupervised (up to 15.66% improvement) and supervised settings (up to 10.94% improvement) on real datasets in terms of NDCG@5. Furthermore, GeoBloom operates up to 80x faster and saves up to 74.72% memory and 87.64% disk space over PLM-based alternatives, rendering it highly potent for real-world applications.

Pre-trained Foundation Models (PFMs) have ushered in a paradigm-shift in AI, due to their ability to learn general-purpose representations that can be readily employed in downstream tasks. While PFMs have been successfully adopted in various fields such as NLP and Computer Vision, their capacity in handling geospatial data remains limited. This can be attributed to the intrinsic heterogeneity of such data, which encompasses different types, including points, segments and regions, as well as multiple information modalities. The proliferation of Volunteered Geographic Information initiatives, like OpenStreetMap, unveils a promising opportunity to bridge this gap. In this paper, we present CityFM, a self-supervised framework to train a foundation model within a selected geographical area. CityFM relies solely on open data from OSM, and produces multimodal representations, incorporating spatial, visual, and textual information. We analyse the entity representations generated by our foundation models from a qualitative perspective, and conduct experiments on road, building, and region-level downstream tasks. In all the experiments, CityFM achieves performance superior to, or on par with, application-specific algorithms.

The prosperity of crowdsourcing geospatial data provides increasing opportunities to understand our cities. In particular, OpenStreetMap (OSM) has become a prominent vault of geospatial data on the Web. In this context, learning urban region representations from OSM data, which is unexplored in previous work, could be profitable for various downstream tasks. In this work, we utilize OSM buildings (footprints) complemented with points of interest (POIs) to learn region representations, as buildings' shapes, spatial distributions, and properties have tight linkages to different urban functions. However, appealing as it seems, urban buildings often exhibit complex patterns to form dense or sparse areas, which brings significant challenges for unsupervised feature extraction. To address the challenges, we propose RegionDCL1, an unsupervised framework to deeply mine urban buildings. In a nutshell, we leverage random points generated by Poisson Disk Sampling to tackle data-sparse areas and utilize triplet loss with a novel adaptive margin to preserve inter-region correlations. Furthermore, we train our model with group-level and region-level contrastive learning, making it adaptive to varying region partitions. Extensive experiments in two global cities demonstrate that RegionDCL consistently outperforms the state-of-the-art counterparts across different region partitions, and outputs effective representations for inferring urban land use and population density.

Many successful methods have been proposed for learning low-dimensional representations on large-scale networks, while almost all existing methods are designed in inseparable processes, learning embeddings for entire networks even when only a small proportion of nodes are of interest. This leads to great inconvenience, especially on large-scale or dynamic networks, where these methods become almost impossible to implement. In this paper, we formalize the problem of separated matrix factorization, based on which we elaborate a novel objective function that preserves both local and global information. We compare our SMF framework with approximate SVD algorithms and demonstrate SMF can capture more information when factorizing a given matrix. We further propose SepNE, a simple and flexible network embedding algorithm which independently learns representations for different subsets of nodes in separated processes. By implementing separability, our algorithm reduces the redundant efforts to embed irrelevant nodes, yielding scalability to large networks. To further incorporate complex information into SepNE, we discuss several methods that can be used to leverage high-order proximities in large networks. We demonstrate the effectiveness of SepNE on several real-world networks with different scales and subjects. With comparable accuracy, our approach significantly outperforms state-of-the-art baselines in running times on large networks.

Heterogeneous information networks (HINs) are ideal for describing real-world data with different types of entities and relationships. To carry out machine learning on HINs, meta-paths are widely utilized to extract semantics with pre-defined patterns, and models such as graph convolutional networks (GCNs) are thus enabled. However, previous works generally assume a fixed set of meta-paths, which is unrealistic as real-world data are overwhelmingly diverse. Therefore, it is appealing if meta-paths can be automatically selected given an HIN, yet existing works aiming at such problem possess drawbacks, such as poor efficiency and ignoring feature heterogeneity. To address these drawbacks, we propose GraphMSE, an efficient heterogeneous GCN combined with automatic meta-path selection. Specifically, we design highly efficient meta-path sampling techniques, and then injectively project sampled meta-path instances to vectors. We then design a novel semantic feature space alignment, aiming to align the meta-path instance vectors and hence facilitate meta-path selection. Extensive experiments on real-world datasets demonstrate that GraphMSE outperforms state-of-the-art counterparts, figures out important meta-paths, and is dramatically (e.g. 200 times) more efficient.

Network embedding is a method of learning a low-dimensional vector representation of network vertices under the condition of preserving different types of network properties. Previous studies mainly focus on preserving structural information of vertices at a particular scale, like neighbor information or community information, but cannot preserve the hierarchical community structure, which would enable the network to be easily analyzed at various scales. Inspired by the hierarchical structure of galaxies, we propose the Galaxy Network Embedding (GNE) model, which formulates an optimization problem with spherical constraints to describe the hierarchical community structure preserving network embedding. More specifically, we present an approach of embedding communities into a low-dimensional spherical surface, the center of which represents the parent community they belong to. Our experiments reveal that the representations from GNE preserve the hierarchical community structure and show advantages in several applications such as vertex multi-class classification, network visualization, and link prediction. The source code of GNE is available online.

Graph Convolutional Networks (GCNs) achieved tremendous success by effectively gathering local features for nodes. However, commonly do GCNs focus more on node features but less on graph structures within the neighborhood, especially higher-order structural patterns. However, such local structural patterns are shown to be indicative of node properties in numerous fields. In addition, it is not just single patterns, but the distribution over all these patterns matter, because networks are complex and the neighborhood of each node consists of a mixture of various nodes and structural patterns. Correspondingly, in this paper, we propose Graph Structural topic Neural Network, abbreviated GraphSTONE 1, a GCN model that utilizes topic models of graphs, such that the structural topics capture indicative graph structures broadly from a probabilistic aspect rather than merely a few structures. Specifically, we build topic models upon graphs using anonymous walks and Graph Anchor LDA, an LDA variant that selects significant structural patterns first, so as to alleviate the complexity and generate structural topics efficiently. In addition, we design multi-view GCNs to unify node features and structural topic features and utilize structural topics to guide the aggregation. We evaluate our model through both quantitative and qualitative experiments, where our model exhibits promising performance, high efficiency, and clear interpretability.