Geoinformatica最新文献

Building a real-time spatio-temporal forecasting system is a challenging problem with many practical applications such as traffic and road network management. Most forecasting research focuses on achieving (often marginal) improvements in evaluation metrics such as MAE/MAPE on static benchmark datasets, with less attention paid to building practical pipelines which achieve timely and accurate forecasts when the network is under heavy load. Transport authorities also need to leverage dynamic data sources such as roadworks and vehicle-level flow data, while also supporting ad-hoc inference workloads at low cost. Our cloud-based forecasting solution Foresight, developed in collaboration with Transport for the West Midlands (TfWM), is able to ingest, aggregate and process streamed traffic data, enhanced with dynamic vehicle-level flow and urban event information, to produce regularly scheduled forecasts with high accuracy. In this work, we extend Foresight with several novel enhancements, into a new system which we term Foresight Plus. New features include an efficient method for extending the forecasting scale, enabling predictions further into the future. We also augment the inference architecture with a new, fully serverless design which offers a more cost-effective solution and which seamlessly handles sporadic inference workloads over multiple forecasting scales. We observe that Graph Neural Network (GNN) forecasting models are robust to extensions of the forecasting scale, achieving consistent performance up to 48 hours ahead. This is in contrast to the 1 hour forecasting periods popularly considered in this context. Further, our serverless inference solution is shown to be more cost-effective than provisioned alternatives in corresponding use-cases. We identify the optimal memory configuration of serverless resources to achieve an attractive cost-to-performance ratio.

Trajectory similarity search is one of the most fundamental tasks in spatial-temporal data analysis. Classical methods are based on predefined trajectory similarity measures, consuming high time and space costs. To accelerate similarity computation, some deep metric learning methods have recently been proposed to approximate predefined measures based on the learned representation of trajectories. However, instead of predefined measures, real applications may require personalized measures, which cannot be effectively learned by existing models due to insufficient labels. Thus, this paper proposes a transfer-learning-based model FTL-Traj, which addresses this problem by effectively transferring knowledge from several existing measures as source measures. Particularly, a ProbSparse self-attention-based GRU unit is designed to extract the spatial and structural information of each trajectory. Confronted with diverse source measures, the priority modeling assists the model for the rational ensemble. Then, sparse labels are enriched with rank knowledge and collaboration knowledge via transfer learning. Extensive experiments on two real-world datasets demonstrate the superiority of our model.

Abstract

The process of computing a representative trajectory for a set of raw (or even semantically enriched) trajectories is an attractive solution to minimize several challenges related to trajectory management, like trajectory data integration or trajectory pattern analysis. We identify two main strategies for accomplishing such a process (trajectory data summarization and trajectory data fusion), but we argue that this subject is still an open issue, and we did not find a survey with such a focus. In order to fill this literature gap, this paper presents a survey that analyzes several issues around the two aforementioned strategies, like the type of representative data computed by each approach, the dimensions that are considered by the approach (spatial, temporal, and semantics), the accomplished methods of the proposed processes, and how the process is evaluated. Additionally, we compare these two research areas (trajectory summarization and trajectory fusion) in literature to analyze their relationship. Finally, some open issues related to this subject are also pointed out.

With the spread of GPS-equipped portable devices, Location-Based Services (LBSs) flourished. Some crucial LBSs require real-time processing of moving spatial-keyword queries over moving objects, such as an ambulance seeking for volunteers. The research community proposed solutions for scenarios assuming that either the queries or the queried objects are moving, but solutions are needed assuming that both are moving. This work proposes SkyEye; a model that efficiently processes moving continuous top-k spatial-keyword queries over moving objects in a directed streets network. SkyEye computes queries’ answer sets for time intervals and smartly updates the answer sets based on the recent history. Novel optimization techniques and indexing structures are leveraged to improve SkyEye’s efficiency and scalability. The mathematical foundations of these optimization techniques are thoroughly demonstrated. Finally, extensive experiments showed that SkyEye has significant performance improvements in terms of efficiency, scalability, and accuracy compared to a baseline model.

In recent years, the exploration of knowledge in large-scale human mobility has gained significant attention. In order to achieve a semantic understanding of human behavior and uncover patterns in large-scale human mobility, Named Entity Recognition (NER) is a crucial technology. The rapid advancements in IoT and CPS technologies have led to the collection of massive human mobility data from various sources. Therefore, there’s a need for Cross-domain NER which can transfer entity information from the source domain to automatically identify and classify entities in different target domain texts. In the situation of the data-poor, how could we transfer human mobility knowledge over time and space is particularly significant, therefore this paper proposes an Adaptive Text Sequence Enhancement Module (at-SAM) to help the model enhance the association between entities in sentences in the data-poor target domains. This paper also proposes a Predicted Label-Guided Dual Sequence Aware Information Module (Dual-SAM) to improve the transferability of label information. Experiments were conducted in domains that contain hidden knowledge about human mobility, the results show that this method can transfer task knowledge between multiple different domains in the data-poor scenarios and achieve SOTA performance.

In recent years, there has been an increased interest in understanding and predicting the weather using weather station data with Spatial-Temporal Graph Neural Networks (STGNN). However, it has large prediction errors as a result of the inherent non-linearities and the influence of dynamic spatio-temporal auto-correlation. Using a continuously-varying graph topology chronologically, while embedding domain knowledge to enforce validity, can effectively resolve the issue, but the implementation of such concept constitutes an interdisciplinary challenge for researchers. A Dynamic Graph Former (DGFormer) model is proposed to address this challenge. It combines a topology learner through a deep generative layer with domain knowledge enhancement inserted into the STGNN structure, where the derived physics-guided method allows for an efficient integration with the earth system. For capture of the optimal topology, we merge a node-embedding-based similarity metric learning and the superposition principle as physical assistants into the dynamic graph module. We evaluate our model with a real-world weather dataset on short-term (12 hours) and medium-range (360 hours) prediction tasks. DGFormer achieves outstanding performance with obvious improvements by up to 34.84% at short-term prediction and by up to 23.25% at medium-range prediction compared with the state-of-the-art methods. We also conducted detailed analyses for cities in three regions and visualized the dynamic graph, revealing the characteristics, advantages, and graph visualization of our model.

One major goal of spatial query processing is to mitigate I/O costs and minimize the search space. However, geometric computation can be heavy-duty for spatial queries, in particular for complex geometries such as polygons with many edges based on a vector-based representation. Many past techniques have been provided for spatial partitioning and indexing, which are mainly built on minimal bounding boxes or other approximation methods and are not optimized for reducing geometric computation. In this paper, we propose a novel vector-raster hybrid approach through rasterization, where rich pixel-centric information is preserved to help not only filter out more candidates but also reduce geometry computation load. Based on the hybrid model, we implement four typical spatial queries, which can be generalized for other types of spatial queries. We also propose cost models to estimate the latency for those query types. Our experiments demonstrate that the hybrid model can boost the performance of spatial queries on complex polygons by up to one order of magnitude.

By tapping into the human mobility of the urban rail transit (URT) network to understand the travel demands and characteristics of passengers in the urban space, URT managers are able to obtain more support for decision-making to improve the effectiveness of operation and management, the travel experience of passengers, as well as public safety. However, not all URT networks have sufficient human mobility data (e.g., newly-operated URT networks). It is necessary to provide data support for mining human mobility in data-poor URT networks. Therefore, we propose a method called Meta Long Short-Term Memory Network (Meta-LSTM) for passenger flow prediction at URT stations to provide data support for networks that lack data. The Meta-LSTM is to construct a framework that increases the generalization ability of a long short-term memory network (LSTM) to various passenger flow characteristics by learning passenger flow characteristics from multiple data-rich stations and then applying the learned parameter to data-scarce stations by parameter initialization. The Meta-LSTM is applied to the URT network of Nanning, Hangzhou, and Beijing, China. The experiments on three real-world URT networks demonstrate the effectiveness of our proposed Meta-LSTM over several competitive baseline models. Results also show that our proposed Meta-LSTM has a good generalization ability to various passenger flow characteristics, which can provide a reference for passenger flow prediction in the stations with limited data.

Ridesharing services have been becoming a prominent solution to reduce road traffic congestion and environmental pollution in urban areas. Existing ridesharing services fall apart in ensuring the social comfort of the riders. We formulate a Community aware Ridesharing Group Set (CaRGS) query that satisfies the spatial and social constraints of the riders and finds a set of ridesharing groups with the maximum number of served riders. The CaRGS query utilizes user social data in community levels to ensure user privacy. We show that the problem of finding CaRGS query answer is NP-Hard and propose two heuristic approaches: a hierarchical approach and an iterative approach to evaluate CaRGS queries. We evaluate the effectiveness, efficiency, and accuracy of our solution through extensive experiments using real datasets and present a comparative analysis among the proposed algorithms.