With the rapid increasing availability of information and popularization of mobility devices, trajectories have become more complex in their form. Trajectory data is now high dimensional, and often associated with heterogeneous sources of semantic data, that are called Multiple Aspect Trajectories. The high dimensionality and heterogeneity of these data makes classification a very challenging task both in term of accuracy and in terms of efficiency. The present demo offers a tool, called AUTOMATISE, to support the user in the classification task of multiple aspect trajectories, specifically for extracting and visualizing the movelets, the parts of the trajectory that better discriminate a class. The AUTOMATISE integrates into a unique platform the fragmented approaches available in the literature for multiple aspects trajectories and, in general, for multidimensional sequence classification into a unique web-based and python library system. We illustrate the architecture and the use of the tool for offering both movelets visualization and a complete configuration of classification experimental settings.

Co-clustering is a specific type of clustering that addresses the problem of finding groups of objects without necessarily considering all attributes. This technique has shown to have more consistent results in high-dimensional sparse data than traditional clustering. In trajectory co-clustering, the methods found in the literature have two main limitations: first, the space and time dimensions have to be constrained by user-defined thresholds; second, elements (trajectory points) are clustered ignoring the trajectory sequence, assuming that the points are independent among them. To address the limitations above, we propose a new trajectory co-clustering method for mining semantic trajectory co-clusters. It simultaneously clusters the trajectories and their elements taking into account the order in which they appear. This new method uses the element frequency to identify candidate co-clusters. Besides, it uses an objective cost function that automatically drives the co-clustering process, avoiding the need for constraining dimensions. We evaluate the proposed approach using a real-world publicly available dataset. The experimental results show that our proposal finds frequent and meaningful contiguous sequences revealing mobility patterns, thereby the most relevant elements.

In the last decade, trajectory classification has received significant attention. The vast amount of data generated on social media, the use of sensor networks, IOT devices and other Internet-enabled sources allowed the semantic enrichment of mobility data, making the classification task more challenging. Existing trajectory classification methods have mainly considered space, time and numerical data, ignoring the semantic dimensions. Only recently proposed methods as Movelets and MASTERMovelets can handle all types of dimensions. MASTERMovelets is the only method that automatically discovers the best dimension combination and subtrajectory size for trajectory classification. However, although it outperformed the state-of-the-art in terms of accuracy, MASTERMovelets is computationally expensive and results in a high dimensionality problem, which makes it unfeasible for most real trajectory datasets that contain a big volume of data. To overcome this problem and enable the application of the movelets approach on large datasets, in this paper we propose a new high-performance method for extracting movelets and classifying trajectories, called HiPerMovelets (High-performance Movelets). Experimental results show that HiPerMovelets is 10 times faster than MASTERMovelets, reduces the high-dimensionality problem, is more scalable, and presents a high classification accuracy in all evaluated datasets with both raw and semantic trajectories.

Abstract During the last few years the volumes of the data that synthesize trajectories have expanded to unparalleled quantities. This growth is challenging traditional trajectory analysis approaches and solutions are sought in other domains. In this work, we focus on data compression techniques with the intention to minimize the size of trajectory data, while, at the same time, minimizing the impact on the trajectory analysis methods. To this extent, we evaluate five lossy compression algorithms: Douglas-Peucker (DP), Time Ratio (TR), Speed Based (SP), Time Ratio Speed Based (TR_SP) and Speed Based Time Ratio (SP_TR). The comparison is performed using four distinct real world datasets against six different dynamically assigned thresholds. The effectiveness of the compression is evaluated using classification techniques and similarity measures. The results showed that there is a trade-off between the compression rate and the achieved quality. The is no “best algorithm” for every case and the choice of the proper compression algorithm is an application-dependent process.

A plethora of applications and devices reporting their locations generate massive amounts of spatiotemporal data along with other useful information. These data can form trajectories with sequences...

The increasing popularity of Location-Based Social Networks (LBSNs) and the semantic enrichment of mobility data in several contexts in the last years has led to the generation of large volumes of trajectory data. In contrast to GPS-based trajectories, LBSN and context-aware trajectories are more complex data, having several semantic textual dimensions besides space and time, which may reveal interesting mobility patterns. For instance, people may visit different places or perform different activities depending on the weather conditions. These new semantically rich data, known as multiple-aspect trajectories, pose new challenges in trajectory classification, which is the problem that we address in this paper. Existing methods for trajectory classification cannot deal with the complexity of heterogeneous data dimensions or the sequential aspect that characterizes movement. In this paper we propose MARC, an approach based on attribute embedding and Recurrent Neural Networks (RNNs) for classifying multiple-aspect trajectories, that tackles all trajectory properties: space, time, semantics, and sequence. We highlight that MARC exhibits good performance especially when trajectories are described by several textual/categorical attributes. Experiments performed over four publicly available datasets considering the Trajectory-User Linking (TUL) problem show that MARC outperformed all competitors, with respect to accuracy, precision, recall, and F1-score.

For many years trajectory similarity research has focused on raw trajectories, considering only space and time information. With the trajectory semantic enrichment, emerged the need for similarity measures that support space, time, and semantics. Although some trajectory similarity measures deal with all these dimensions, they consider only stops, ignoring the moves. We claim that, for some applications, the movement between stops is as important as the stops, and they must be considered in the similarity analysis. In this article, we propose SMSM, a novel similarity measure for semantic trajectories that considers both stops and moves. We evaluate SMSM with three trajectory datasets: (i) a synthetic trajectory dataset generated with the Hermoupolis semantic trajectory generator, (ii) a real trajectory dataset from the CRAWDAD project, and (iii) the Geolife dataset. The results show that SMSM overcomes state-of-the-art measures developed either for raw or semantic trajectories.

Trajectory data analysis and mining require distance and similarity measures, and the quality of their results is directly related to those measures. Several similarity measures originally proposed for time-series were adapted to work with trajectory data, but these approaches were developed for well-behaved data that usually do not have the uncertainty and heterogeneity introduced by the sampling process to obtain trajectories. More recently, similarity measures were proposed specifically for trajectory data, but they rely on simplistic movement uncertainty representations, such as linear interpolation. In this article, we propose a new distance function, and a new similarity measure that uses an elliptical representation of trajectories, being more robust to the movement uncertainty caused by the sampling rate and the heterogeneity of this kind of data. Experiments using real data show that our proposal is more accurate and robust than related work.

Unusual behavior detection has been of interest in video analysis, transportation systems, movement trajectories, and so on. In movement trajectories, only a few works identify unusual behavior of objects around pre-defined points of interest (POI), such as surveillance cameras, commercial buildings, etc., that may be interesting for several application domains, mainly for security. In this article, we define new types of unusual behaviors of moving objects in relation to POI, including surround, escape, and return. Based on these types of unusual behavior, we (i) present an algorithm to compute these behaviors, (ii) define a set of functions to weight the degree of unusual behavior of every moving object in the database, and (iii) rank the moving objects according to the degree of unusual behavior in relation to a set of POIs. We evaluate the proposed method with real trajectory data and show that the closest work does not detect the proposed behaviors and ranks objects considering only one type of unusual movement.

Summary form only given. The recent advances and price reduction of technologies for collecting spatial and spatio-temporal data like Satellite Images, Cellular Phones, Sensor Networks, and GPS devices has facilitated the collection of data referenced in space and time. These huge collections of data often hide interesting information which conventional systems and classical data mining techniques are unable to discover. Spatial and spatio-temporal data are embedded in continuous space, whereas classical datasets (e.g. transactions) are often discrete. Spatial and spatio-temporal data require complex data preprocessing, transformation, data mining, and post-processing techniques to extract novel, useful, and understandable patterns. The importance of spatial and spatio-temporal data mining is growing with the increasing incidence and importance of large geo-spatial datasets such as maps, repositories of remote-sensing images, trajectories of moving objects generated by mobile devices, etc. Applications include Mobile-commerce industry (location-based services), climatologically effects of El Nino, land-use classification and global change using satellite imagery, finding crime hot spots, local instability in traffic, migration of birds, fishing control, pedestrian behavior analysis, and so on. Thus, new methods are needed to analyze spatial and spatio-temporal data to extract interesting, useful, and non-trivial patterns. The main goal of this tutorial is to disseminate this research field, giving an overview of the current state of the art and the main methodologies and algorithms for spatial and spatio-temporal data mining. This tutorial is directed to researches and practitioners, experts in data mining, analysts of spatial and spatio-temporal data, as well as knowledge engineers and domain experts from different application areas.

Many association rule‐mining algorithms have been proposed in the last few years. Their main drawback is the huge amount of generated patterns. In spatial association rule mining, besides the large amount of rules, many are well‐known geographic domain associations explicitly represented in geographic database schemas. Existing algorithms have only considered the data, while the schema has not been considered. The result is that also the associations explicitly represented in geographic database schemas are extracted by association rule‐mining algorithms. With the aim to reduce the number of well‐known patterns and association rules, this paper presents a summary of results of a novel approach to extract patterns from geographic databases. A two step‐pruning method is presented to avoid the generation of association rules that are previously known to be uninteresting. Experiments with real geographic databases show a considerable time reduction in both geographic data pre‐processing and spatial association rule mining, with a very significant reduction in the total number of rules.

The collection of moving object data is becoming more and more common, and therefore there is an increasing need for the efficient analysis and knowledge extraction of these data in different application domains. Trajectory data are normally available as sample points, and do not carry semantic information, which is of fundamental importance for the comprehension of these data. Therefore, the analysis of trajectory data becomes expensive from a computational point of view and complex from a user's perspective. Enriching trajectories with semantic geographical information may simplify queries, analysis, and mining of moving object data. In this paper we propose a data preprocessing model to add semantic information to trajectories in order to facilitate trajectory data analysis in different application domains. The model is generic enough to represent the important parts of trajectories that are relevant to the application, not being restricted to one specific application. We present an algorithm to compute the important parts and show that the query complexity for the semantic analysis of trajectories will be significantly reduced with the proposed model.

The large amount of patterns generated by frequent pattern mining algorithms has been extensively addressed in the last few years. In geographic pattern mining, besides the large amount of patterns, many are well known geographic domain associations. Existing algorithms do not warrant the elimination of all well known geographic dependences since no prior knowledge is used for this purpose. This paper presents a two step method for mining frequent geographic patterns without associations that are previously known as non-interesting. In the first step the input space is reduced as much as possible. This is as far as we know still the most efficient method to reduce frequent patterns. In the second step, all remaining geographic dependences that can only be eliminated during the frequent set generation are removed in an efficient way. Experiments show an elimination of more than 50% of the total number of frequent patterns, and which are exactly the less interesting.

In frequent geographic pattern mining a large amount of patterns is well known a priori. This paper presents a novel approach for mining frequent geographic patterns without associations that are previously known as non- interesting. Geographic dependences are eliminated during the frequent set generation using prior knowledge. After the dependence elimination maximal generalized frequent sets are computed to remove redundant frequent sets. Experimental results show a significant reduction of both the number of frequent sets and the computational time for mining maximal frequent geographic patterns.