Search

In order to improve the intelligent effect of English performance rating, this paper analyzes the English performance rating process and builds the corresponding algorithm model based on the wisdom of teaching ideas, and it combines BP spatial network technology with spatial sampling and spatial statistical algorithms to construct the system kernel algorithm. Moreover, this paper extracts students’ English answers through text recognition and compares them with the standard library. Among them, objective questions are mainly scored directly through standard answer comparison, and subjective questions compare student scores with standard answers. In addition, this paper uses a needs analysis to construct the framework of the entire English performance rating system and evaluates the practical effect of the system constructed in the paper through experimental research. According to statistical research data, the English performance rating system in this paper is known to have certain effects.

In the study of human action recognition, two-stream networks have made excellent progress recently. However, there remain challenges in distinguishing similar human actions in videos. This paper proposes a novel local-aware spatio-temporal attention network with multi-stage feature fusion based on compact bilinear pooling for human action recognition. To elaborate, taking two-stream networks as our essential backbones, the spatial network first employs multiple spatial transformer networks in a parallel manner to locate the discriminative regions related to human actions. Then, we perform feature fusion between the local and global features to enhance the human action representation. Furthermore, the output of the spatial network and the temporal information are fused at a particular layer to learn the pixel-wise correspondences. After that, we bring together three outputs to generate the global descriptors of human actions. To verify the efficacy of the proposed approach, comparison experiments are conducted with the traditional hand-engineered IDT algorithms, the classical machine learning methods (i.e., SVM) and the state-of-the-art deep learning methods (i.e., spatio-temporal multiplier networks). According to the results, our approach is reported to obtain the best performance among existing works, with the accuracy of 95.3% and 72.9% on UCF101 and HMDB51, respectively. The experimental results thus demonstrate the superiority and significance of the proposed architecture in solving the task of human action recognition.

In many applications of graph processing, the input data is often generated from an underlying geometric point data set. However, existing high-performance graph processing frameworks assume that the input data is given as a graph. Therefore, to use these frameworks, the user must write or use external programs based on computational geometry algorithms to convert their point data set to a graph, which requires more programming effort and can also lead to performance degradation. In this paper, we present our ongoing work on the Geo- Graph framework for shared-memory multicore machines, which seamlessly supports routines for parallel geometric graph construction and parallel graph processing within the same environment. GeoGraph supports graph construction based on k-nearest neighbors, Delaunay triangulation, and b-skeleton graphs. It can then pass these generated graphs to over 25 graph algorithms. GeoGraph contains highperformance parallel primitives and algorithms implemented in C++, and includes a Python interface. We present four examples of using GeoGraph, and some experimental results showing good parallel speedups and improvements over the Higra library. We conclude with a vision of future directions for research in bridging graph and geometric data processing.

Toxicity analysis is a major challenge in drug design and discovery. Recently significant progress has been made through machine learning due to its accuracy, efficiency, and lower cost. US Toxicology in the 21st Century (Tox21) screened a large library of compounds, including approximately 12 000 environmental chemicals and drugs, for different mechanisms responsible for eliciting toxic effects. The Tox21 Data Challenge offered a platform to evaluate different computational methods for toxicity predictions. Inspired by the success of multiscale weighted colored graph (MWCG) theory in protein—ligand binding affinity predictions, we consider MWCG theory for toxicity analysis. In the present work, we develop a geometric graph learning toxicity (GGL-Tox) model by integrating MWCG features and the gradient boosting decision tree (GBDT) algorithm. The benchmark tests of the Tox21 Data Challenge are employed to demonstrate the utility and usefulness of the proposed GGL-Tox model. An extensive comparison with other state-of-the-art models indicates that GGL-Tox is an accurate and efficient model for toxicity analysis and prediction.

Automated characterization of spatial data is a kind of critical geographical intelligence. As an emerging technique for characterization, spatial Representation Learning (SRL) uses deep neural networks (DNNs) to learn non-linear embedded features of spatial data for characterization. However, SRL extracts features by internal layers of DNNs, and thus suffers from lacking semantic labels. Texts of spatial entities, on the other hand, provide semantic understanding of latent feature labels, but is insensible to deep SRL models. How can we teach a SRL model to discover appropriate topic labels in texts and pair learned features with the labels? This paper formulates a new problem: feature-topic pairing, and proposes a novel Particle Swarm Optimization (PSO) based deep learning framework. Specifically, we formulate the feature-topic pairing problem into an automated alignment task between 1) a latent embedding feature space and 2) a textual semantic topic space. We decompose the alignment of the two spaces into: 1) point-wise alignment, denoting the correlation between a topic distribution and an embedding vector; 2) pair-wise alignment, denoting the consistency between a feature-feature similarity matrix and a topic-topic similarity matrix. We design a PSO based solver to simultaneously select an optimal set of topics and learn corresponding features based on the selected topics. We develop a closed loop algorithm to iterate between 1) minimizing losses of representation reconstruction and feature-topic alignment and 2) searching the best topics. Finally, we present extensive experiments to demonstrate the enhanced performance of our method.

FOSS4G 2021 Buenos Aires, Buenos Aires, Argentine, 27-/09/2021 - 02/10/2021; Arabesque is an application for the exploration and geovisualisation of origin-destination flows (or spatial networks), developed within the framework of the Univ. Gustave Eiffel (ex. IFSTTAR)-funded research project geographic flow visualisation (gflowiz) geoflowiz, in collaboration with the CNRS. It allows both the exploration and the filtering of OD data and their representation, with a strong emphasis on geographic information layering and features' semiology. The key-objective is to propose an easy way to produce a modern cartography (a geovisualisation) of thematic flows (e.g. bilateral flow volume), at several geographic scales, even from your own datasets. The objective of this article is to position Arabesque in the range of geoweb applications for producing flow maps, by comparing its functionalities with those of similar web applications- Magrit, Kepler.gl, flowmap.blue -pointing out their respective advantages and limitations. The analysis of its functionalities is compared on the same flow dataset -MOBSCO, i.e. a dataset describing the school mobility of French pupils and students on a given year- for a practical and empirical "validation" of its contributions. We demonstrate that the configurations and appearances of these tools' visual output depend largely on the culture of their developers, and on the use and audiences for which they have been developed.

Colocation mining is useful for understanding the interactions or dependencies that occur among geographic phenomena. Most colocation mining methods are based on planar space. However, in urban spaces, many human-related activities are constrained by a road network. Planar colocation mining methods are not suitable for studying the concerning geographic phenomena in an urban space. In this paper, we propose a visualization method to discover colocation patterns constrained by a road network. The method consists of two major components: network kernel density estimation and network colocation rule map construction. In the colocation rule map construction component, spatial interactions among spatial network geographic phenomena are modeled based on the idea of color mixing. We use simulated datasets with different spatial patterns, different sample sizes, and different maximum distances between road network events to test our method. The results show that our method is effective for mining colocation patterns in different situations. We also change the resolution of the network colocation rule maps, and the results show that the resolution has little influence on the results. In the case study, we apply our method to explore the spatial association between crimes and city facilities in the Loop and the Near North Side districts of Chicago.

In real world, cooperation among unrelated agents remains to represent one open challenge in many disciplines. In some cases, subjects can even change their social dilemmas to ensure their own benefit. Inspired by this fact, here we propose a multigame (composed of prisoner's dilemma game and snowdrift game) and its coevolution mechanism in networked populations: if a player succeeds to study the strategy of its opponent, it also learns the game type of its opponent. Based on numerous computation simulations, it is unveiled that compared with the setup of multigame, coevolution mechanism can effectively resolve the problem of collective cooperation. While for this observation, it is attributed to the fact that vast majority of players changed their games to snowdrift game (i.e. lower social dilemma), where cooperation-defection bistable state guarantees cooperation clusters. We hope our finding can inspire more studies for resolving the social dilemmas.

Automatic liver and tumor segmentation remain a challenging topic, which subjects to the exploration of 2D and 3D contexts in CT volume. Existing methods are either only focus on the 2D context by treating the CT volume as many independent image slices (but ignore the useful temporal information between adjacent slices), or just explore the 3D context lied in many little voxels (but damage the spatial detail in each slice). These factors lead an inadequate context exploration together for automatic liver and tumor segmentation. In this paper, we propose a novel full-context convolution neural network to bridge the gap between 2D and 3D contexts. The proposed network can utilize the temporal information along the Z axis in CT volume while retaining the spatial detail in each slice. Specifically, a 2D spatial network for intra-slice features extraction and a 3D temporal network for inter-slice features extraction are proposed separately and then are guided by the squeeze-and-excitation layer that allows the flow of 2D context and 3D temporal information. To address the severe class imbalance issue in the CT volume and meanwhile improve the segmentation performance, a loss function consisting of weighted cross-entropy and jaccard distance is proposed. During the network training, the 2D and 3D contexts are learned jointly in an end-to-end way. The proposed network achieves competitive results on the Liver Tumor Segmentation Challenge (LiTS) and the 3D-IRCADB datasets. This method should be a new promising paradigm to explore the contexts for liver and tumor segmentation.

The urban structure is the spatial reflection of various economic and cultural factors acting on the urban territory. Different from the physical structure, urban structure is closely related to the population mobility. Taxi trajectories are widely distributed, completely spontaneous, closely related to travel needs, and massive in data volume. Mining it not only can help us better understand the flow pattern of a city, but also provides a new perspective for interpreting the urban structure. On the basis of massive taxi trajectory data in Chengdu, we introduce a network science approach to analysis, propose a new framework for interaction analysis, and model the intrinsic connections within cities. The spatial grid of fine particles and the trajectory connections between them are used to resolve the urban structure. The results show that: (1) Based on 200,000 taxi trajectories, we constructed a spatial network of traffic flow using the interaction analysis framework and extracted the cold hot spots among them. (2) We divide the 400 traffic flow network nodes into 6 communities. Community 2 has high centrality and density, and belongs to the core built-up area of the city. (3) A traffic direction field is proposed to describe the direction of the traffic flow network, and the direction of traffic flow roughly presents an inflow from northeast to southwest and an outflow from southeast to northwest of the study area. The interaction analysis framework proposed in this study can be applied to other cities or other research areas (e.g., population migration), and it could extract the directional nature of the network as well as the hierarchical structure of the city.

Space Syntax and the theory of natural movement demonstrated that spatial morphology is a primary factor influencing movement. This paper investigates to what extent spatial morphology at different scales (node, community and global network) influences the use of public space by micromobility. An axial map and corresponding network for Lisbon’s walkable and open public space, and data from e-scooters parking locations, is used as case study. Relevant metrics and their correlations (intelligibility, accessibility, permeability and local dimension) for the quantitative characterization of spatial morphology properties are described and computed for Lisbon’s axial map. Communities are identified based on the network topological structure in order to investigate how these properties are affected at different scales in the case study. The resulting axial line clustering is compared via the variation of information metric with the clustering obtained from e-scooters’ proximity. The results obtained enable to conclude that the space syntax properties are scale dependent in Lisbon’s pedestrian network. On the other hand both the correlation between these properties, the number of scooters and the variation of information between clusters indicate that the spatial morphology is not the only factor influencing micromobility. Through the comparative analysis between the main properties of the public space network of Lisbon and data collected from e-scooters locations in a timeframe, centrality becomes a dynamic concept, relying not only on the static topological properties of the urban network, but also on other quantitative and qualitative factors, since the flows’ operating on the network will operate several transformations on the spatial network properties through time, uncovering spatiotemporal dynamics.

Transportation networks allow us to model flows of people and resources across geographic space, but the people and resources we wish to model are often not natively tied to our networks. Instead, they can occur as point data (such as store, train station, and domicile locations) and/or grid data (such as socio-economic and aggregate area data). Here we present a set of methods to integrate point data into an augmented transportation network. This method facilitates analyses of temporo-spatial measures (such as accessibility scores) using only efficient breadth first search algorithms. We demonstrate the approach by calculating walkability scores for the train stations within the central Tokyo area.

Given a collection of multi-attribute trajectories, an event definition, and a spatial network, the Significant Lagrangian Linear Hotspot Discovery (SLLHD) problem finds the paths where records in the trajectories tend to be events in the Lagrangian perspective. The SLLHD problem is of significant societal importance because of its applications in transportation planning, vehicle design, and environmental protection. Its main challenges include the potentially large number of candidate hotspots caused by the tremendous volume of trajectories as well as the non-monotonicity of the statistic measuring event concentration. The related work on the linear hotspot discovery problem is designed in the Eulerian perspective and focuses on point datasets, which ignores the dependence of event occurrence on trajectories and the paths where trajectories are. To solve this problem, we introduce an algorithm in the Lagrangian perspective, as well as five refinements that improve its computational scalability. Two case studies on real-world datasets and experiments on synthetic data show that the proposed approach finds hotspots which are not detectable by existing techniques. Cost analysis and experimental results on synthetic data show that the proposed approach yields substantial computational savings.

We present a new measure, CMetric, to classify driver behaviors using centrality functions. Our formulation combines concepts from computational graph theory and social traffic psychology to quantify and classify the behavior of human drivers. CMetric is used to compute the probability of a vehicle executing a driving style, as well as the intensity used to execute the style. Our approach is designed for realtime autonomous driving applications, where the trajectory of each vehicle or road-agent is extracted from a video. We compute a dynamic geometric graph (DGG) based on the positions and proximity of the road-agents and centrality functions corresponding to closeness and degree. These functions are used to compute the CMetric based on style likelihood and style intensity estimates. Our approach is general and makes no assumption about traffic density, heterogeneity, or how driving behaviors change over time. We present an algorithm to compute CMetric and demonstrate its performance on real-world traffic datasets. To test the accuracy of CMetric, we introduce a new evaluation protocol (called "Time Deviation Error") that measures the difference between human prediction and the prediction made by CMetric., Comment: Accepted to IROS 2020

In order to effectively evaluate personnel performance, a distributed data mining algorithm for spatial networks based on BP neural wireless network is proposed. In the cloud computing environment, an excavator is used to construct multiple input multiple output spatial network data, analyze the data structure, and perform redundant data compression of massive data through time-frequency feature extraction. Combined with adaptive matching filtering method, the characteristics of the data are matched. The spatial frequency feature extraction method is used to locate the features of the multiple-input multiple-output spatial network data. In order to improve the accuracy of data mining, the BP neural network is used to classify and identify the extracted data features to achieve the optimization of data mining. A wireless sensor network is a wireless network composed of a large number of stationary or moving sensors in a self-organizing and multi-hop manner. It cooperatively senses, collects, processes, and transmits the information of the perceived objects in the geographical area covered by the network, and finally puts these The information is sent to the owner of the network. This algorithm improves the accuracy of personnel performance evaluation, and simultaneously establishes a hierarchical analysis and quantitative evaluation model for the performance of government managers, and adjusts the results of hierarchical statistical analysis on government administrators as needed. The performance evaluation and optimization of government administrators were introduced. The empirical analysis results show that the method has higher accuracy for government managers' performance evaluation, higher efficiency of big data processing and better integration.

Spatial Design Network Analysis (sDNA) is a toolbox for 3-d spatial network analysis, especially street/path/urban network analysis, motivated by a need to use network links as the principal unit of analysis in order to analyse existing network data. sDNA is usable from QGIS & ArcGIS geographic information systems, AutoCAD, the command line, and via its own Python API. It computes measures of accessibility (reach, mean distance/closeness centrality, gravity), flows (bidirectional betweenness centrality) and efficiency (circuity) as well as convex hull properties, localised within lower- and upper-bounded radial bands. Weighting is flexible and can make use of geometric properties, data attached to links, zones, matrices or combinations of the above. Motivated by a desire to base network analysis on route choice and spatial cognition, the definition of distance can be network-Euclidean, angular, a mixture of both, custom, or specific to cyclists (avoiding slope and motorised traffic). In addition to statistics on network links, the following outputs can be computed: geodesics, network buffers, accessibility maps, convex hulls, flow bundles and skim matrices. Further tools assist with network preparation and calibration of network models to observed data.To date, sDNA has been used mainly for urban network analysis both by academics and city planners/engineers, for tasks including prediction of pedestrian, cyclist, vehicle and metro flows and mode choice; also quantification of the built environment for epidemiology and urban planning & design.

Recent years have witnessed the effective of attention network based on two-stream for video action recognition. However, most methods adopt the same structure on spatial stream and temporal stream, which produce amount redundant information and often ignore the relevance among channels. In this paper, we propose a channel-wise spatial attention with spatiotemporal heterogeneous framework, a new approach to action recognition. First, we employ two different network structures for spatial stream and temporal stream to improve the performance of action recognition. Then, we design a channel-wise network and spatial network inspired by self-attention mechanism to obtain the fine-grained and salient information of the video. Finally, the feature of video for action recognition is generated by end-to-end training. Experimental results on the datasets HMDB51 and UCF101 shows our method can effectively recognize the actions in the video.

Many real-world spatial systems can be conceptualized as networks. In these conceptualizations, nodes and links represent system components and their interactions, respectively. Traditional network analysis applies graph theory measures to static network datasets. However, recent interest lies in the representation and analysis of evolving networks. Existing network automata approaches simulate evolving network structures, but do not consider the representation of evolving networks embedded in geographic space nor integrating actual geospatial data. Therefore, the objective of this study is to integrate network automata with geographic information systems (GIS) to develop a novel modelling framework, Geographic Network Automata (GNA), for representing and analyzing complex dynamic spatial systems as evolving geospatial networks. The GNA framework is implemented and presented for two case studies including a spatial network representation of (1) Conway&rsquo, s Game of Life model and (2) Schelling&rsquo, s model of segregation. The simulated evolving spatial network structures are measured using graph theory. Obtained results demonstrate that the integration of concepts from geographic information science, complex systems, and network theory offers new means to represent and analyze complex spatial systems. The presented GNA modelling framework is both general and flexible, useful for modelling a variety of real geospatial phenomena and characterizing and exploring network structure, dynamics, and evolution of real spatial systems. The proposed GNA modelling framework fits within the larger framework of geographic automata systems (GAS) alongside cellular automata and agent-based modelling.

Spatial network analysis is a collection of methods for measuring accessibility potentials as well as for analyzing flows over transport networks. Though it has been part of the practice of geographic information systems for a long time, designing network analytical workflows still requires a considerable amount of expertise. In principle, artificial intelligence methods for workflow synthesis could be used to automate this task. This would improve the (re)usability of analytic resources. However, though underlying graph algorithms are well understood, we still lack a conceptual model that captures the required methodological know-how. The reason is that in practice this know-how goes beyond graph theory to a significant extent. In this article we suggest interpreting spatial networks in terms of quantified relations between spatial objects, where both the objects themselves and their relations can be quantified in an extensive or an intensive manner. Using this model, it becomes possible to effectively organize data sources and network functions towards common analytical goals for answering questions. We tested our model on 12 analytical tasks, and evaluated automatically synthesized workflows with network experts. Results show that standard data models are insufficient for answering questions, and that our model adds information crucial for understanding spatial network functionality.

Non-spatial ecological networks provide insight into the organization and interaction between biological entities. More recently, biological dispersal is modelled using spatial networks, static sets of georeferenced habitat patches that connect based on a species’ maximum dispersal distance. However, dispersal is complex, where spatial patterns at the landscape scale emerge from interactions between ecological entities and landscape features at much finer individual scales. Agent-based modelling (ABM) is a computational representation of complex systems capable of capturing this complexity. Therefore, this study develops a network-ABM (N-ABM) that combines network and complex systems theory to simulate complex evolving spatial networks. The developed N-ABM approach is implemented on the case study of the emerald ash borer (EAB) bark beetle using geospatial datasets in Ontario, Canada. The N-ABM generates dynamic spatial network structures that emerge from interactions between the EAB and tree agents at the individual scale. The resulting networks are analyzed using graph theory measures. Analysis of the results indicates a relationship between preferential attachment in insect host selection and the emergent scale-free network structure. The N-ABM approach can be used to represent dynamic ecological networks and provides insight into how network structure emerges from EAB dispersal dynamics, useful for forest management.

The relationship between cross-regional cultural landscapes does not currently receive enough attention in cultural landscape conservation. Cultural landscapes in China are faced with the crisis of fragmentation and islanding, which makes it necessary to strengthen the idea of integrated conservation. The Great Wall cultural belt in Beijing is a cross-regional linear cultural landscape in the northwest mountainous area of Beijing. With the Great Wall as a link connecting the surrounding natural, social, and cultural resources to form a banded structure, it has the typical and practical significance of integrated conservation. Based on the theory of the cultural landscape security pattern and its ‘node – setting – connection’ model, this research studies the spatial structure and shaping mechanism of the Great Wall cultural belt in Beijing from the perspective of tourist flow, using Flickr geo-tagged photos and Wikiloc tracks. The results identify the core nodes and intermediate nodes of the Great Wall cultural belt in Beijing and show that it has a multi-core, multi-group hierarchical nesting structure differing from the geographic space, which has groups following the evolution orbit of ‘linear string – circumnuclear star – complete network.’ To explain the differences between the cultural association network (CAN) and the geospatial network (GSN), we find that the cultural associations between nodes rely on geophysical channels, including heritage corridors and traffic passages, as physical carriers, but the nodes’ attraction and their setting elements, containing natural geography environments, settlements, administrative divisions, etc., change the channel resistance, so CAN and GSN present different characteristics. From the perspective of systematic and spatial network thoughts, this study has made an attempt to adopt new analysis and research methods to achieve the integrated conservation and inheritance of linear cultural landscapes in mountainous areas.

1. Pollinators, such as bees, face the complex challenge of efficiently exploiting patchily distributed floral resources across large landscapes. 2. In the present study we consider the utility of spatial network statistics to analyse the foraging patterns of bees moving between feeding sites at various spatial and temporal scales. 3. We explain how spatial movement networks can be derived theoretically and experimentally to describe bee foraging decisions. 4. We illustrate this approach by analysing six datasets of bumblebees and honeybees foraging in arrays of artificial flowers, and showing how network metrics change as foragers gain experience with the spatial distribution of feeding sites. 5. We compare network analyses with more conventional statistics used to characterise bee foraging movements and discuss the implications of our novel statistical and modelling approach for pollination ecology.

Latest research in the field of fully autonomous driving led to a new control approach based on decentralized control algorithms for macroscopic traffic scenarios. This control approach balances the traffic load (density) within a road network segment of limited area to maximize the vehicle throughput and to avoid traffic jams. However, it is necessary to divide the road network into multiple individual segments of limited area in an appropriate way. Due to the lack of a methodology to locate such locations for limited areas within a road network, this work presents an adopted technique that solves this problem by modelling the road network as spatial network and by using graph centrality indices. The presented method was implemented into a software framework and validated by simulation.

Given a spatial network and a collection of activities (e.g., pedestrian fatality reports, crime reports), Significant Linear Hotspot Discovery (SLHD) finds all shortest paths in the spatial network where the concentration of activities is statistically significantly high. SLHD is important for societal applications in transportation safety or public safety such as finding paths with significant concentrations of accidents or crimes. SLHD is challenging because 1) there are a potentially large number of candidate paths ( $\sim10^{16}$ ) in a given dataset with millions of activities and road network nodes and 2) test statistic (e.g., density ratio) is not monotonic. Hotspot detection approaches on euclidean space (e.g., SaTScan) may miss significant paths since a large fraction of an area bounded by shapes in euclidean space for activities on a path will be empty. Previous network-based approaches consider only paths between road intersections but not activities. This paper proposes novel models and algorithms for discovering statistically significant linear hotspots using the algorithms of neighbor node filter, shortest path tree pruning, and Monte Carlo speedup. We present case studies comparing the proposed approaches with existing techniques on real data. Experimental results show that the proposed algorithms yield substantial computational savings without reducing result quality.

Task functional magnetic resonance imaging (fMRI) has been widely employed for brain activation detection and brain network analysis. Modeling rich information from spatially-organized collection of fMRI time series is challenging because of the intrinsic complexity. Hypothesis-driven methods, such as the general linear model (GLM), which regress exterior stimulus from voxel-wise functional brain activity, are limited due to overlooking the complexity of brain activities and the diversity of concurrent brain networks. Recently, sparse representation and dictionary learning methods have attracted increasing interests in task fMRI data analysis. The major advantage of this methodology is its promise in reconstructing concurrent brain networks systematically. However, this data-driven strategy is, to some extent, arbitrary and does not sufficiently utilize the prior information of task design and neuroscience knowledge. To bridge this gap, we here propose a novel supervised sparse representation and dictionary learning framework based on stochastic coordinate coding (SCC) algorithm for task fMRI data analysis, in which certain brain networks are learned with known information such as pre-defined temporal patterns and spatial network patterns, and at the same time other networks are learned automatically from data. Our proposed method has been applied to two independent task fMRI datasets, and qualitative and quantitative evaluations have shown that our method provides a new and effective framework for task fMRI data analysis.

Traffic forecasting is an important prerequisite for the application of intelligent transportation systems in urban traffic networks. The existing works adopted RNN and CNN/GCN, among which GCRN is the state-of-the-art work, to characterize the temporal and spatial correlation of traffic flows. However, it is hard to apply GCRN to the large-scale road networks due to high computational complexity. To address this problem, we propose abstracting the road network into a geometric graph and building a Fast Graph Convolution Recurrent Neural Network (FastGCRNN) to model the spatial-temporal dependencies of traffic flow. Specifically, we use FastGCN unit to efficiently capture the topological relationship between the roads and the surrounding roads in the graph with reducing the computational complexity through importance sampling, combine GRU unit to capture the temporal dependency of traffic flow, and embed the spatiotemporal features into Seq2Seq based on the Encoder-Decoder framework. Experiments on large-scale traffic data sets illustrate that the proposed method can greatly reduce computational complexity and memory consumption while maintaining relatively high accuracy.

Road networks are a type of spatial network, where edges may be associated with qualitative information such as road type and speed limit. Unfortunately, such information is often incomplete; for instance, OpenStreetMap only has speed limits for 13% of all Danish road segments. This is problematic for analysis tasks that rely on such information for machine learning. To enable machine learning in such circumstances, one may consider the application of network embedding methods to extract structural information from the network. However, these methods have so far mostly been used in the context of social networks, which differ significantly from road networks in terms of, e.g., node degree and level of homophily (which are key to the performance of many network embedding methods). We analyze the use of network embedding methods, specifically node2vec, for learning road segment embeddings in road networks. Due to the often limited availability of information on other relevant road characteristics, the analysis focuses on leveraging the spatial network structure. Our results suggest that network embedding methods can indeed be used for deriving relevant network features (that may, e.g, be used for predicting speed limits), but that the qualities of the embeddings differ from embeddings for social networks., Best Paper at the 3rd IEEE International Workshop on Big Spatial Data (BSD 2018)

Spatial hotspot discovery aims at discovering regions with statistically significant concentration of activities. It has shown great value in many important societal applications such as transportation engineering, public health, and public safety. This paper formulates the problem of Linear Hotspot Detection on All Simple Paths (LHDA) which identifies hotspots from the complete set of simple paths enumerated from a given spatial network. LHDA overcomes the limitations of existing methods which miss hotspots that naturally occur along linear simple paths on a road network. To address the computational challenges, we propose a novel algorithm named bidirectional fragment-multi-graph traversal (ASP_FMGT) and two path reduction approaches ASP_NR and ASP_HD. Experimental analyses show that ASP_FMGT has substantially improved performance over state-of-the-art approach (ASP_Base) while keeping the solution complete and correct. Moreover, a case study on real-world datasets showed that ASP_FMGT outperforms existing approaches.

A hotsport is a reflection of dense areas of human activities in a city. Hotspots can be widely used in public applications, such as transport management, disease prevention and control, crowd management, environmental monitoring. Nevertheless, existing work mainly utilize static models to detect hotspots in spatial dimension, which ignores the evolving characteristic of hotspots in temporal dimension. Such work does not help to study the evolution law of man-land relationships behind the hotspots in real-life applications. In this paper, we propose a dynamic spatial network (DSN) considering evolving characteristics to discover hotspots (collective gathering, traffic congestion, and so on) in urban road networks. Based on the model, we propose a generate-and-test framework for hotspot detection, which generates the (k+1)-edge candidate hotspots based on the k-edge hotspots, and then tests each candidate if it is qualified and closed. Finally, experiments on real taxi data have verified the effectiveness and efficiency of the proposed concepts and algorithms.

This paper describes the use of graph analysis for spatial networks. The use of community detection algorithms for detecting communities- groups of similar objects within networks of land cover objects to determine the land use is evaluated. Land cover to land use transformation requires some knowledge to merge land cover objects. Community detection algorithms merge objects of the formed spatial network into communities. Community detection algorithms are efficient analysis tool for spatial graphs and can identify land use communities but with different characteristics, although spatial networks with topological relationships between objects can cause some problems.

Presently a days remarkable improvement of the web innovation and sensor innovation, the worldwide spatial information, alongside its huge volume and multi-scale highlights, clarifies higher prerequisites for information association. The spatial network has been generally utilized in many guide list. In any case, when spatial articles have multi-scale, the spatial matrix list with proportionate networks has been critized for its low-efficiencies. To tackle this issue, here proposed an intelligent list just as upset record dependent on GeoSOT subdivision lattices. By methods for this list, we can accomplish question task productively paying little heed to the sizes of spatial articles.

The geometric graphs miners found on the literature are very inefficient.GeoSuM is able to accurately find frequent geometric subgraphs.The used growth approaches greatly reduces the search space.The geometric tolerance can be reached by using a geometric similarity function.The are many practical applications for the geometric graphs miners. There are many problems where graphs are involved; some of them use geometric information. In the former context, the geometric subgraph mining algorithms have many applications, since resulting patterns provide relevant information for areas like biometrics, computer vision or molecular classification. Despite this fact, in the literature, there are only a few of geometric graph miners, and most of them have restrictions reducing the availability for application on real environments. In this paper, a new geometric subgraph mining algorithm is proposed. This algorithm, named GeoSuM, uses a pattern growth approach based on the already known non-geometric graph miner, gSpan. The proposal is very efficient in comparison with other state of the art approaches, in terms of execution time of the mining process. GeoSuM was tested in two different applications: molecular classification and fingerprint matching. Experimental results prove the flexibility and impact of GeoSuM in different scenarios.

In citation networks, the content relativity of papers is a precondition of engendering citations, which is hard to model by a topological graph. A geometric graph is proposed to predict some features of the citation networks with exponentially growing papers, which addresses the precondition by using coordinates of nodes to model the research contents of papers, and geometric distances between nodes to diversities of research contents between papers. Citations between modeled papers are drawn according to a geometric rule, which addresses the precondition as well as some other factors engendering citations, namely academic influences of papers, aging of those influences, and incomplete copying of references. Instead of cumulative advantage of degree, the model illustrates that the scale-free property of modeled networks arises from the inhomogeneous academic influences of modeled papers. The model can also reproduce some other statistical features of citation networks, e.g. in- and out-assortativities, which show the model provides a suitable tool to understand some aspects of citation networks by geometry.

The efficient visualization of dynamic network structures has become a dominant problem in many big data applications, such as large network analytics, traffic management, resource allocation graphs, logistics, social networks, and large document repositories. In this paper, we present a large-graph visualization system called ModuleGraph. ModuleGraph is a scalable representation of graph structures by treating a graph as a set of modules. The main objectives are: (1) to detect graph patterns in the visualization of large-graph data, and (2) to emphasize the interconnecting structures to detect potential interactions between local modules. Our first contribution is a hybrid modularity measure. This measure partitions the cohesion of the graph at various levels of details. We aggregate clusters of nodes and edges into several modules to reduce the overlap between graph components on a 2D display. Our second contribution is a k-clustering method that can flexibly detect the local patterns or substructures in modules. Patterns of modules are preserved by the ModuleGraph system to avoid information loss, while sub-graphs are clustered as a single node. Our experiments show that this method can efficiently support large-scale social and spatial network visualization. Graphical Abstract text

The aim of mining spatial co-location patterns is to find the corresponding subsets of spatial features that have strong spatial correlation in the real world. This is an important technology for the extraction and comprehension of implicit knowledge in large spatial databases. However, existing methods of co-location mining consider events as taking place in a homogeneous and isotropic context in Euclidean space, whereas the physical movement in an urban space is usually constrained by a road network. Furthermore, previous works do not take the ‘distance decay effect’ of spatial interactions into account, which may reduce the effectiveness of the result. Here we propose an improved spatial co-location pattern mining method, including the network-constrained neighborhood and addition of a distance-decay function, to find the spatial dependence between network phenomena e.g. urban facilities. The underlying idea is to utilize a model function in the interest measure calculation to weight the contribution of a co-location to the overall interest measure instance inversely proportional to the separation distance. Our approach was evaluated through extensive experiments using facility points-of-interest data sets. The results show that the network-constrained approach is a more effective method than the traditional one in network-structured space. The proposed approach can also be applied to other human activities e.g. traffic accidents constrained by a street network.

A new approach for mining spatial co-location patterns was presented.The neighborhood was refined using network distances rather than Euclidean ones.An efficient algorithm to build the neighborhood relationship graph was proposed.The performance of the proposed algorithm was explored. With the evolution of geographic information capture and the emergency of volunteered geographic information, it is getting more important to extract spatial knowledge automatically from large spatial datasets. Spatial co-location patterns represent the subsets of spatial features whose objects are often located in close geographic proximity. Such pattern is one of the most important concepts for geographic context awareness of location-based services (LBS). In the literature, most existing methods of co-location mining are used for events taking place in a homogeneous and isotropic space with distance expressed as Euclidean, while the physical movement in LBS is usually constrained by a road network. As a result, the interestingness value of co-location patterns involving network-constrained events cannot be accurately computed. In this paper, we propose a different method for co-location mining with network configurations of the geographical space considered. First, we define the network model with linear referencing and refine the neighborhood of traditional methods using network distances rather than Euclidean ones. Then, considering that the co-location mining in networks suffers from expensive spatial-join operation, we propose an efficient way to find all neighboring object pairs for generating clique instances. By comparison with the previous approaches based on Euclidean distance, this approach can be applied to accurately calculate the probability of occurrence of a spatial co-location on a network. Our experimental results from real and synthetic data sets show that the proposed approach is efficient and effective in identifying co-location patterns which actually rely on a network.

Virtual water flows are used to determine the indirect water requirements of a region or product, making them an indispensable tool for water sustainability analysis and assessment. Commodity flows are a key data needed to compute virtual water but are typically available every 5 years in the United States (US). The lack of continuous, annual commodity flow data severely limits our ability to study and understand the drivers, evolution, and alterations of virtual water in the US. We build and evaluate a machine learning model using Random Forest (RF) to predict annual commodity and virtual water flow networks. The model is used to perform several modeling experiments and illustrate the prediction of annual virtual water flows in the US during 2013–2018. We show that the RF predictions consistently outperform those from a gravity model. The overall performance of the RF algorithm improves as commodities or regions are aggregated into coarser groups. Likewise, the inclusion of past commodity flows as an additional explanatory variable enhances the RF performance. The combination of RF classification and regression allows predicting both network connections and flows without comprising performance. Based on our RF predictions for 2013–2018, we find that temporal variations in virtual water flows can be large for some regions in the US, underscoring the need addressed by this study of reconstructing domestic virtual water changes over time. By capturing inter-regional water consumption interactions in space and time, such reconstructed data could be beneficial in the future for anticipating and managing local and regional water scarcity.

The storage resources and communication resources of spatial network nodes are generally very limited, and they have the characteristics of large time delay and cannot guarantee the existence of end-to-end path all the time. Messages may reside in the intermediate nodes for a long time, which makes the congestion of spatial network inevitable and leads to the decline of network performance. In order to solve this problem, an effective congestion control scheme for space delay/disruption networks based on early unloading is proposed. In this scheme, the change rate of node cache is monitored at all times, and measures are taken in advance when congestion is about to occur, so that messages are transmitted through the non-optimal path prior to the optimal path, so as to alleviate the node storage pressure. Simulation results indicate that the presented scheme can achieve higher success rate of message delivery and more uniform network traffic distribution.

This study aims at developing a non-(semi-)parametric method to extract the hidden network structure from the {0,1}-valued distribution flow data with missing observations on the links between nodes. Such an input data type widely exists in the studies of information propagation process, such as the rumor spreading through social media. In that case, a social network does exist as the media of the spreading process, but its link structure is completely unobservable; therefore, it is important to make inference of the structure (links) of the hidden network. Unlike the previous studies on this topic which only consider abstract networks, we believe that apart from the link structure, different social-economic features and different geographic locations of nodes can also play critical roles in shaping the spreading process, which has to be taken into account. To uncover the hidden link structure and its dependence on the external social-economic features of the node set, a multidimensional spatial social network model is constructed in this study with the spatial dimension large enough to account for all influential social-economic factors. Based on the spatial network, we propose a nonparametric mean-field equation to govern the rumor spreading process and apply the likelihood estimator to make inference of the unknown link structure from the observed rumor distribution flows. Our method turns out easily extendible to cover the class of block networks that are useful in most real applications. The method is tested through simulated data and demonstrated on a data set of rumor spreading on Twitter.

Crowd flow analysis is a important part of urban computing, which is playing an vital role in urban plan and management. This paper addresses the challenges in current population flow analysis. We propose a user space dynamic discovery method based on graph signal model. Taking the base station as a spatial node, the spatial dependence relationship between the nodes is modeled as a spatial network map. The spectral wavelet operator is applied to the spatial map signal to generate wavelet coefficients on different wavelet scales. The simulation results, show that our scheme can be used for valuable information such as the origin, spread and span of population mobility.

With massive real-world networks added with spatial attributes, query processing in spatial networks has been widely used in real-life applications. Considering the high costs of spatial network queries, outsourcing querying services to cloud provides a cost-effective way for data owners. However, directly outsourcing such services may cause serious privacy concerns. To address this privacy problem, existing studies apply anonymous or differential privacy techniques on spatial network queries, which are not under the semi-trusted secure model. Other existing works primarily focus on query processing over either encrypted spatial or network data, which cannot directly be applied to work out the secure spatial network query problem. To this end, we define and study Secure Spatial Network kNN Query (SSNQ) problem on cloud platform. We first present Basic Secure Spatial Network kNN Query (BSSNQ) method, in which we compute secure kNN for the query node to construct candidate sequences using secure subprotocols. With pre-encrypted network distances, we then compute shortest paths between the query node and each candidate, and securely update the candidate sequences according to Euclidean restriction to derive final query results. To improve the efficiency of BSSNQ, we further propose Heuristic Secure Spatial Network kNN Query (HSSNQ) method, which securely calculates shortest paths from the query node to the visiting nodes iteratively, and use optimistic estimate distances as the key to lead the heuristic search within a priority queue. Thorough analysis shows the security and complexity of the proposed methods, and extensive experimental results on real datasets demonstrate the efficiency of the query performance.

Due to the recent popularization of the Geographic Information System (GIS), spatial network environments that can display the changes of spatial axes on mobile phones are receiving great attention. In spatial network environments, since a query object that seeks location information selects several candidate target objects based on the search conditions, we often use a k-nearest neighbor (kNN) search, which seeks several target objects near the query object. However, since a kNN search needs to find the kNN by calculating the distance from the query to all the objects, the computational complexity might become too large, depending on the number of objects. To reduce this computation time in a kNN search, many researchers have proposed a search method that divides regions using a Voronoi diagram. In this method, we reduce the computational complexity by generating Voronoi regions for each object based on the positional relationships between two objects. In addition, a search method was proposed to reduce the processing time by generating Voronoi diagrams using a contact zone. However, since conventional methods generate Voronoi diagrams for objects in order, the computational complexity might become too large, depending on the number of objects. In this paper, we propose a generation method of the Voronoi diagram to the reduce processing time by parallelizing the generation of Voronoi regions using a contact zone. In our evaluation, we confirmed that the processing time under the proposed method was reduced about 15.9% more than conventional methods that are not parallelized.

This paper presents a solution to the problem posed in the ACM SIGSPATIAL GIS Cup 2018, namely to identify all upstream features---nodes and edges existing on simple paths between starting points and controllers---in a given spatial network. Our approach is based primarily on a two-sweep depth-first search which decomposes a graph into its biconnected components prior to collecting the upstream features. Our algorithm runs in linear time in the size of the graph and, in practice, is able to solve large instances with millions of features within seconds on an ordinary personal laptop, identifying features if and only if they are upstream.

Almost all spatial systems can be modelled as networks. Typically, networks representations of real systems are static, useful for generating descriptive network measures. However, recent interest lies in the representation and analysis of evolving spatial networks that can facilitate the examination of the close coupling between network structure and network dynamics. Therefore, this study proposes a novel spatial modelling framework, Geographic Network Automata (GNA), to represent evolving spatial networks as a function of iteratively applied transition rules. The framework is demonstrated using an adapted and spatially explicit version of Conway's Game of Life. The simulated spatial network structures are quantified using network measures. The presented GNA model framework is general and flexible so that different types of geospatial phenomena can be modelled, providing valuable insights into the link between network structure, dynamics, and evolution.