Spatiotemporal event sequences represent the sequences of event types whose spatiotemporal instances frequently follow each other in spatiotemporal context. In this work, we present spatiotemporal event sequence mining from spatio-temporal event datasets that contains evolving region trajectories. We propose two algorithms for discovering spatio-temporal event sequences. We formally define a flexible spatiotemporal follow relationship, introduce various data models for capturing the sequence forming behavior. Lastly, we present an extended experimental evaluation that demonstrates the computational efficiency of our algorithms.
This paper focuses on the study of detecting real-time pedestrian flow by using Bluetooth on smartphones to show evacuation route. When a natural disaster strikes, knowing real-time accurate pedestrian density and flow in wide range is very important to lead people to the safe places. Evacuation route can be calculated by agent-based simulation and the real-time pedestrian flow. One promising way to detect real-time pedestrian density and flow is mobile sensing. With devices the evacuee have, the pedestrian density of near area can be detected. In this paper, we introduce the method of detecting pedestrian density and propose how to apply this method to a disaster for solving crowded situation.
Currently there are demands for maximization of taxi services and also for saving fuel usage within massive cities. Spatial big data extracted from taxi service records and GPS can be used to suggest optimal routing options to achieve these goals. The taxi cab ride data contains 7,000 unique taxies being serviced in Seoul, South Korea. In this study one week worth of data with the size of 3.13GB were used. Also road network data provided by Ministry of Land, Infrastructure and Transport (MOLIT), which contains 19,229 nodes and 22,192 links, and census map provided by Statistics Korea were used as base-map. Lastly floating population data of Seoul city area, gathered with mobile phones, has been used as an index of demand for taxi service. By using taxi cab ride data, which contains trajectory with time and 2D coordinates, and information about whether passenger is on the taxi or not, hot spots were analyzed for 1) taxies without passengers whom are available to pick-up passengers, 2) places where people are experiencing difficulty hailing a taxi due to high demand for taxi. Combination of these two types of hot spots can provide new insight for both public and commercial sectors to maximize the efficiency of taxi service and to reduce idle fuel usage. Afterwards the floating population data is used to provide indices for taxi usage in Seoul area, providing further insights. Utilizing the time stamp records on the taxi GPS data, hourly based hot spots for both 'demand' and 'supply' for taxi cab ride can be derived, and this outcome can be practically used to guide taxi drivers to high demanding places and avoid high supplying places.
Citizens in various locations can report local infrastructure issues to the government by posting reports on certain websites, such as Chiba Report and FixMyStreet. Recently, these systems have begun operating worldwide. In these systems, a large volume of information is collected on infrastructure problems that are identified by citizens (e.g., broken paving slabs, fly tipping, graffiti, potholes). This information is expected to be utilized for infrastructure maintenance. However, local problems (especially road defection) identified by citizens are sometimes not deemed an urgent matter for road managers. This is because it is difficult for an average person to determine road damage status. Furthermore, non-critical reports may be a burden for local government because each report requires visual confirmation. We therefore propose a smartphone application based on a deep neural network that can determine road damage status using only photographs of the road. This application is based on a deep neural network model trained by citizen reports and road manager inspection results, which are gathered daily on a government server. The application updates the model parameters each time it launches and thereby becomes increasingly more intelligent and effective. The proposed system enables average citizens to easily determine road damage status using only a smartphone application. In addition, because not only expert road managers but also local government officials without expert knowledge can inspect the road, the proposed system can be useful for local governments that lack expert road managers.
This paper describes a new Network-constrained Moving objects indexing structure, which extends the state-of-the-art for this kind of data. The indexing structure we propose is called Temporally Enhanced Network-Constrained R-tree (TENC R-tree), which solves the shortcomings in other Network-Constrained access methods like the FNR-tree [7], MON-tree [1] and UTR-tree. These existing indexing methods are designed to store and retrieve the moving objects based on spatial features, followed by their temporal ones. They are generally not efficient when a query has only temporal constraints, or when a specific moving object id is also part of the query conditions. In such cases, existing methods have to scan the entire database to retrieve the result. Furthermore, the aforementioned methods are not efficient in processing Strict-path query, which is a query type that retrieves trajectories that follow all the edges in the queried path [10]. Our proposed TENC R-tree index allows good performance for almost all types of queries on moving objects in a constrained network, whether the constraints are spatial, temporal, or based on object id. Also, the TENC R-tree out-performs other access methods on the case of Path queries. Our experiments show the performance has been improved by 10 to 100 times for such queries.
Trajectory analysis is a central problem in the era of big data due to numerous interconnected mobile devices generating unprecedented amounts of spatio-temporal trajectories. Unfortunately, datasets of spatial trajectories are quite difficult to analyse because of the computational complexity of the various existing distance measures. A significant amount of work in comparing two trajectories stems from calculating temporal alignments of the involved spatial points. With this paper, we propose an alignment-free method of representing spatial trajectories using low-dimensional feature vectors by summarizing the combinatorics of shape-derived string sequences. Therefore, we propose to translate trajectories into strings describing the evolving shape of each trajectory, and then provide a sparse matrix representation of these strings using frequencies of adjacencies of characters (n-grams). The final feature vectors are constructed by approximating this matrix with low-dimensional column space using singular value decomposition. New trajectories can be projected into this geometry for comparison. We show that this construction leads to low-dimensional feature vectors with surprising expressive power. We illustrate the usefulness of this approach in various datasets.
Advancement in mobile and GPS technologies have enabled users to record and publish their route activities or trajectories through location based social networking sites. Existing research mainly focus on finding popular routes and recommending suitable routes for the users based on the historical movements of users between different Point of Interests (POIs). However, users often spend most of their time around different POIs (e.g., Colosseo) and less time traveling between POIs. Thus, existing methods fail to capture the detailed movement of users around a POI, which we call Region of Interest (ROI). A major challenge of identifying patterns of routes inside an ROI comes from the inaccurate and incomplete data of user trajectories. In this paper we propose a novel technique to find the most popular path within an ROI from historical trajectory data by rephrasing trajectories into smaller part and eliminating noisy points from trajectories. We then devise an algorithm to produce the most popular path inside each ROI. We perform experiments on a real dataset extracted from Flickr to show the effectiveness of our approach.
We put forth a system, to predict distant-future positions of multiple moving entities and index the forecasted trajectories, in order to answer predictive queries involving long time horizons. Today, the proliferation of mobile devices with GPS functionality and internet connectivity has led to a rapid development of location-based services, accounting for user mobility prediction as a key paradigm. Mobility prediction is already playing a major role in traffic management, urban planning and location-based advertising, which demand accurate and long time horizon forecasting of user movements. Existing prediction methodologies either use motion patterns or techniques based on frequently visited places for predicting the next move. However, when it comes to distant-future, human mobility is too complex to be represented by such statistical functions. Therefore, the existing techniques are not well suited to answer distant-future queries with a satisfactory level of accuracy. To tackle this problem, we introduce a novel spatial object, 'Representative Trajectory', which embodies the movements of users amongst their zones of interest. We propose means to empirically evaluate the quality of this object and dynamically adapt its extraction method based on user mobility behaviour. We rely on an inverted index to store the predicted trajectories that scales well with the number of moving entities. Our evaluation results show that the technique achieves more than 70% accurate predictions with the best extraction technique. This shows that longer query time horizons do not necessarily demand complex spatial indexing schemes, which have to be rebalanced as they grow and which is a constantly experienced problem while answering predictive queries.
Many indoor positioning methods and systems exhibit high inaccuracies and structural positioning biases, when deployed and evaluated in real-world environments. This holds especially for signal-strength-based positioning, the prevalent means for position tracking in environments, that are not suitable for GNSS positioning, such as large building complexes. In such environments though strong positioning inaccuracies and biases result from the many building elements with different attenuation properties. We propose and evaluate deviation maps as a means for capturing, and thereby reducing, positioning errors and biases as prevalent in different parts of deployment's building complex.
Given a sequence S of temporally ordered observations, non necessarily of spatial nature, the segmentation task partitions S in a set of disjoint sub-sequences s i , .., s n - the segments - such that ∪ i ∈[1, n ] s i = S . Typically, segments represents sub-sequences that are somehow homogeneous with respect to some criteria. Depending on the context and the nature of observations, segments can be given an approximated representation, for example segments can be assigned a descriptive label or one of the data points is chosen as representative of the whole sub-sequence. The final result is a summarized representation of the sequence. This simple and intuitive mechanism has been extensively studied in literature, for example, for the summarization of time series. Interestingly, the notion of segment is also at the basis of the most recent trajectory data models. For example, segments are the informative units in the semantic trajectories, where they are called episodes. Episodes are spatial sub-trajectories that can be semantically annotated using application-dependent descriptions, e.g. place names [1]. Similarly the recent symbolic trajectory data model [2] describes the individual movement as a sequence of temporally annotated labeled states s 1 , .. s n , where each state s i is associated with a time interval. Beyond data modeling, segmentation can be employed for the indexing of trajectories in moving object databases while another major role is to support data analysis, especially for the extraction of individual mobility patterns. The concept of trajectory segment is thus emerging as shared and perhaps unifying concept across data modeling, indexing and analysis.
We study five existing map construction algorithms, designed and tested with urban vehicle data in mind, and apply them to hiking trajectories with different terrain characteristics. Our main goal is to better understand the existing strategies and their limitations, in order to shed new light into the current challenges for map construction algorithms. We carefully analyze the results obtained by each algorithm focusing on the local details of the generated maps. Our analysis includes the characterization of 10 types of common artifacts, which occur in the results of more than one algorithm, and 7 algorithmic-specific artifacts, which are consequences of different algorithmic strategies. This allows us to extract systematic conclusions about the main challenges to fully automatize the construction of maps from trajectory data, to detect the strengths and weaknesses of the potential different strategies, and to suggest possible ways to design higher-quality map construction methods. We consider that this analysis will be of help for designing new and better methods that perform well in wider and more realistic contexts, not only for road map or hiking reconstruction, but also for other types of trajectory data.
This paper presents a system architecture of a web GIS that is used to develop a web mapping app for real-time macroeconomic impact decision support tool. It incorporates web GIS on the cloud with an autonomous software system for real-time situational awareness (outage statue and economic loss) from power & electric utilities. Our web GIS is a system of systems, and we deployed ESRI's ArcGIS platform, Amazon Web Services (AWS), enterprise spatial database, C#, RESTful API, and JSON format. The system implementation results in a web GIS that contains a GIS server with a set of REST APIs of GIS web services (map service, geodata service, etc) on the cloud that can be used by web mapping apps, mobile GIS apps, or desktop programs to share, display, analyze, and update a geodatabase, which is embedded in cloud. To evaluate our approach, we developed a web map application and an operations dashboard that used the created GIS web services and APIs. Our web GIS is applicable for the "Internet of Things" domain, public safety, cloud communication, crisis response, web map application, location-based services, and real-time GIS.
Trajectory acquisition, management, and processing are important tasks for any application that deals with spatiotemporal data. In order to perform these tasks effectively, it is important to rely on flexible structures. Many data models have been proposed for representing spatiotemporal traces. However, modeling trajectory characteristics and context information is still a challenge. In this work, we introduce the STEP ontology (Semantic Trajectory Episodes) for trajectory enrichment. In order to model this domain, we structure trajectories and related contextual data in terms of semantic episodes that allow describing various characteristics of the traces and context along time and space dimensions. We demonstrate the usage of the STEP ontology for enriching raw trajectories and show how the proposed model may be useful for trajectory analysis tasks.
We are under the big data microscope, and our digital traces are an inestimable source of awareness to deeply understand mobility phenomena as well as economic trends, social relationships and so on. Setting the focus of the big data microscope to capture human systematic behavior is surely a promising direction. The proposed vision is a methodological framework aimed to deal with intelligent personal data store that are able to automatically perform individual data mining, and that can provide proactive suggestions and support decisions, allow to share individual profiles in order to reach a level of knowledge comparable to those belonged to a collective system, and suggest interactions between individual and collective data mining in order to overtake the level of complex society knowledge extracted by the state-of-art methods. The study of individuals profiles, and the comparison and interactions with collective patterns, is dramatically helpful both for the novel detailed information retrieved through the methodological framework and for the possibility to deal at the same time with privacy issues.
Public safety requires emergency response that is timely and efficient. This paper describes how to distribute the emergency call among those in the area so as to optimize their locations when called, and their expertise. The call will route to the next most qualified if those in the immediate vicinity cannot come. We describe the system architecture and provide the algorithmic rules, as well as sketch an interface and propose how the completed system could be evaluated.
The detection of stay-jump-and-moving movement episodes using only cellular data is a big challenge due to the nature of the data. In this article, we propose a method to automatically detect the movement episodes (stay-jump-and-moving) from sparsely sampled spatio-temporal data, in our case Call Detail Records (CDRs), using switching Kalman filter with a new integrated movement model and cellular coverage optimization approach. The algorithm is capable of estimating the movement episodes and classifying the trajectory sequences associated to a stay, a jump or a moving action. The result of this approach can be beneficial for applications using cellular data related to traffic management, mobility profiling, and semantic enrichment.
Concept Geo-tagging is the process of assigning a textual identifier that describes a real-world entity to a physical geographic location. A concept can either be a spatial concept where it possesses a spatial presence or be a non-spatial concept where it has no explicit spatial presence. Geo-tagging locations with non-spatial concepts that have no direct relation is a very useful and important operation but is also very challenging. The reason is that, being a non-spatial concept, e.g., crime, makes it hard to geo-tag it. This paper proposes using the semantic information associated with concepts and locations such as the type as a mean for identifying these relations. The co-occurrence of spatial and non-spatial concepts within the same textual resources, e.g., in the web, can be an indicator of a relationship between these spatial and non-spatial concepts. Techniques are presented for learning and modeling relations among spatial and non-spatial concepts from web textual resources. Co-occurring concepts are extracted and modeled as a graph of relations. This graph is used to infer the location types related to a concept. A location type can be a hospital, restaurant, an educational facility and so forth. Due to the immense number of relations that are generated from the extraction process, a semantically-guided query processing algorithm is introduced to prune the graph to the most relevant set of related concepts. For each concept, a set of most relevant types are matched against the location types. Experiments evaluate the proposed algorithm based on its filtering efficiency and the relevance of the discovered relationships. Performance results illustrate how semantically-guided query processing can outperform the baseline in terms of efficiency and relevancy. The proposed approach achieves an average precision of 74% across three different datasets.
Several different models have been defined in literature for the definition of 3D city models, from CityGML [14] to Inspire [8]. Such models include a geometrical representation of features together with a semantical classification of them. The semantical characterization of objects encapsulates important meaning and relations which are defined only implicitly or through natural language, such as a window surface shall be contained in the building boundary. The problem of ensuring the coherence between geometric and semantic information is well known in literature. Many attempts exist which try to extent the OCL language in order to represent spatial constraints for an UML model. However, this approach requires a deep knowledge of the OCL language and the implementation of ad-hoc procedures for the validation of the defined constraints. The aim of this paper is the development of a set of templates for expressing spatial 3D constraints between features which does not require any particular knowledge of a formal language. Moreover, the constraints instantiated from these templates can be automatically translated into validation procedures.
Advances in mobile and sensor technologies have enabled the collection of continuously changing data such as locations and weather measurements. However, conserving the energy of these devices has been a major challenge. In this work, we propose an energy-efficient solution to a new variant of the Discrete Unit Disk Covering Problem (DUDC), which models a mobile sensor network. We present an approximation algorithm for this problem and theoretical analysis in the case of randomly positioned sensors that shows that three objectives are met: reduce the average number of active sensors that report measurements, spread the measurement burden over time evenly among the reporting sensors and maintain an acceptable quality of the reported measurements. Experimental and theoretical results show that our proposed algorithm has computational complexity and approximation factor comparable to currently known deterministic algorithms while meeting the aforementioned objectives.
Many applications of high societal relevance -- e.g., transportation and traffic management, disaster remediation, location-aware social networking, (tourist) recommendation systems, military logistics (to name but a few) -- rely on some kind of Location Based Services (LBS). The crucial components to support such services, in turn, rely on efficient techniques for managing the data capturing the information pertaining to the whereabouts in time of the moving entities -- storing, retrieving and querying such data. Traditionally, such topics were subjects of the fields called Spatial/Spatio-Temporal Databases, Moving Objects Databases (MOD) and Geographic Information Systems (GIS) [2, 5, 11]. To give an intuitive idea about the magnitude -- according to Mc Kinsey survey from 2011 [9], the volume of location-in-time data exceeds the order of Peta-Bytes per year just from smartphones -- and this is only the "pure" GPS (Global Positioning System) data. Including the cell-towers location data would boost the size by two orders of magnitude -- however, this is not even close to the full magnitude of the variety of location-related data contained in numerous tweets and other social networks based communications (which is of interest for applications such as behavioral marketing).
