Segmentation techniques for the summarization of individual mobility data

Acquaviva

Proceedings of the 28th International Conference on Advances in Geographic Information Systems

et al. 2020

Self Cite

In this paper, we investigate the suitability of state-of-the-art representation learning methods to the analysis of behavioral similarity of moving individuals, based on CDR trajectories. The core of the contribution is a novel methodological framework, mob2vec, centered on the combined use of a recent symbolic trajectory segmentation method for the removal of noise, a novel trajectory generalization method incorporating behavioral information, and an unsupervised technique for the learning of vector representations from sequential data. mob2vec is the result of an empirical study conducted on real CDR data through an extensive experimentation. As a result, it is shown that mob2vec generates vector representations of CDR trajectories in low dimensional spaces which preserve the similarity of the mobility behavior of individuals.

Section: Seqscan-dmentioning

confidence: 99%

Learning Behavioral Representations of Human Mobility

Acquaviva

Proceedings of the 28th International Conference on Advances in Geographic Information Systems

et al. 2020

Self Cite

“…Such a parameter 28 is, however, hard to set, especially whenever the sampling intervals are irregular and the clustering is to be applied to a large number of trajectories. As a result, these techniques are highly time consuming and not very effective in practice [17]. A first attempt to deal in a more systematic way with the problem of noise in clustering-based segmentation is represented by SeqScan [14].…”

Section: Location Discovery Techniquesmentioning

confidence: 99%

“…In literature, the trajectories of concern are commonly of spatial type, thus consisting of coordinated points, while the problem is formulated in terms of trajectory segmentation, namely to find the sub-trajectories (segments) satisfying certain conditions. Two major classes of solutions are the attributecentric and pattern-centric segmentation techniques [17]. Attribute-centric are called the methods partitioning a spatial trajectory into a minimum number of segments in such a way that the movement inside each segment is nearly uniform with respect to some condition on movement attributes.…”

Section: Location Discovery Techniquesmentioning

confidence: 99%

“…SeqScan identifies two classes of outliers: the points indicating a temporary absence from the cluster (local noise) and the points representing a definitive departure from the cluster towards another cluster (transition points). The segmentation algorithm determines the sequence of clusters along with the classified outliers [17].…”

Section: Location Discovery Techniquesmentioning

confidence: 99%

See 1 more Smart Citation

On Location Relevance and Diversity in Human Mobility Data

ACM Trans. Spatial Algorithms Syst.

Quadri

et al. 2020

Self Cite

The theme of human mobility is transversal to multiple fields of study and applications, from ad-hoc networks to smart cities, from transportation planning to recommendation systems on social networks. Despite the considerable efforts made by a few scientific communities and the relevant results obtained so far, there are still many issues only partially solved, that ask for general and quantitative methodologies to be addressed. A prominent aspect of scientific and practical relevance is how to characterize the mobility behavior of individuals. In this article, we look at the problem from a location-centric perspective: we investigate methods to extract, classify and quantify the symbolic locations specified in telco trajectories, and use such measures to feature user mobility. A major contribution is a novel trajectory summarization technique for the extraction of the locations of interest, i.e. attractive, from symbolic trajectories. The method is built on a density-based trajectory segmentation technique tailored to telco data, which is proven to be robust against noise. To inspect the nature of those locations, we combine the two dimensions of location attractiveness and frequency into a novel location taxonomy, which allows for a more accurate classification of the visited places. Another major contribution is the selection of suitable entropy-based metrics for the characterization of single trajectories, based on the diversity of the locations of interest. All these components are integrated in a framework utilized for the analysis of 100,000+ telco trajectories. The experiments show how the framework manages to dramatically reduce data complexity, provide high-quality information on the mobility behavior of people and finally succeed in grasping the nature of the locations visited by individuals.

“…Stop-and-move detection. A number of techniques for the detection of stops and places have become quite popular in recent times (Damiani and Hachem, 2017). A pioneering technique is the CB-SMoT algorithm (Palma et al, 2008).…”

Section: Related Workmentioning

confidence: 99%

Cluster-based trajectory segmentation with local noise

Issa

et al. 2018

Data Min Knowl Disc

Self Cite

We present a framework for the partitioning of a spatial trajectory in a sequence of segments based on spatial density and temporal criteria. The result is a set of temporally separated clusters interleaved by sub-sequences of unclustered points. A major novelty is the proposal of an outlier or noise model based on the distinction between intra-cluster (local noise) and inter-cluster noise (transition): the local noise models the temporary absence from a residence while the transition the definitive departure towards a next residence. We analyze in detail the properties of the model and present a comprehensive solution for the extraction of temporally ordered clusters. The effectiveness of the solution is evaluated first qualitatively and next quantitatively by contrasting the segmentation with ground truth. The ground truth consists of a set of trajectories of labeled points simulating animal movement. Moreover, we show that the approach can streamline the discovery of additional derived patterns, by presenting a novel technique for the analysis of periodic movement. From a methodological perspective, a valuable aspect of this research is that it combines the theoretical investigation with the application and external validation of the segmentation framework. This paves the way to an effective deployment of the solution in broad and challenging fields such as e-science.