Journal Title
Title of Journal: Soc Netw Anal Min
|
Abbravation: Social Network Analysis and Mining
|
Publisher
Springer Vienna
|
|
|
|
Authors: Riccardo Guidotti Anna Monreale Salvatore Rinzivillo Dino Pedreschi Fosca Giannotti
Publish Date: 2016/08/12
Volume: 6, Issue: 1, Pages: 59-
Abstract
The availability of massive digital traces of individuals is offering a series of novel insights on the understanding of patterns characterizing human mobility Many studies try to semantically enrich mobility data with annotations about human activities However these approaches either focus on places with high frequencies eg home and work or relay on background knowledge eg public available points of interest In this paper we depart from the concept of frequency and we focus on a high level representation of mobility using network analytics The visits of each driver to each systematic destination are modeled as links in a bipartite network where a set of nodes represents drivers and the other set represents places We extract such network from two real datasets of human mobility based respectively on GPS and GSM data We introduce the concept of mobility complexity of drivers and places as a ranking analysis over the nodes of these networks In addition by means of community discovery analysis we differentiate subgroups of drivers and places according both to their homogeneity and to their mobility complexityOne of the most fascinating challenges of our time is to understand the complexity of the global interconnected society and possibly to predict human behavior A great part of human behavior is observable through individual movements registered in many different layers mobile phone network GPS devices social media applications road sensors credit card transactions etc Movement is the “hardware” of our daily life We move to perform any activity we have to move to bring children at school to buy a new electronic device to meet with colleagues at work etc If we understand the patterns of human movement we can also comprehend the mechanics of human behaviorOn the basis of this assumption in the last years we have witnessed many studies exploring movements data to understand different aspects related to the mobility of individuals such as the density of traffic Giannotti et al 2011 the identification of systematic movements Trasarti et al 2011 the identification of groups of drivers following common routes Monreale et al 2009 and many others On one hand the movement is an objective phenomenon that can be observed measured and recorded easily with the modern ICT services On the other hand the intended activity of each movement is not always easy to sense and register A common approach to better understand movement behavior consists into the study of the motivations that push an individual to move toward a given destination There are proposals in the literature to semantically enrich movement data on the basis of movement dynamics and properties For example Jiang et al 2012 tries to estimate home/work locations of an individual by analyzing the frequency she visits a particular place Lafferty et al 2001 observe a sequence of movements to derive the sequence of activities performed Rinzivillo et al 2014 extract a series of individual mobility network to learn structured patterns of visits to places and Furletti et al 2013 exploit the background knowledge of the points of interest POIs available in a territory to derive the activities of persons stopping nearbyIn this paper we propose an approach that can be considered as an intermediate step between the movement dynamics exploration and the semantic enrichment of movements We start from the analysis of individual movements to understand the relevance of each destination However we are not interested in the specific activity a person is performing on her destination rather we focus on the “relevance” that a specific destination has for the personA wellknown proverb says that “Home is where the Hearth is” meaning that the home for an individual is not just a mere geographical place but it represents a complex mixture of sensations perceptions and feelings linked to that place It goes without saying that this kind of definition is strongly tied to a personal and subjective vision of that place From the analytical point of view it is difficult to measure this perception The approaches based on semantic enrichment are focused either on places of general interest like restaurants shopping center or on individualbased destinations like home or work Our proposal tries to fill this gap by starting from an individual ranking of personal places to generalize to collective relevance of destinationsConcretely we propose an approach based on complex network analytics methods to model the relevance of a place p according to the persons visiting p The basic intuition is based on the concept of complexity of individual mobility a person d is complex if she visits many different complex places In a similar way a place p has a high relevance ie it is complex if it is visited by many complex visitors This interwined relation among users and places is modeled by means of a bipartite graph called Drivers–Places network Starting from this model we propose two analytical processes based on ranking measures and community discovery In the first process we try to understand both the mobility complexity of people moving in a territory and the mobility complexity of places for the collectivity Therefore the analysis is focused on the mobility behavior of drivers with respect to some specific places which are considered important for both their individual mobility and the collective mobility and on the mobility in the interesting places with respect to the drivers who visit them In the second analytical process based on application of community discovery algorithms we characterize the groups of similar drivers and places with respect to mobility complexityWe experiment our analytical methodology in real case studies considering both GSM and GPS datasets of trajectories Our finding is that drivers and places complexity in terms of mobility can be characterized according to the similarity of the movements that lead a certain user in a certain location Then by doing a deeper analysis with GPS data we show how certain communities are characterized by their topological structure and by their mobility Finally as additional point studying ranking measures we demonstrate that the method we use to calculate the mobility complexity scores is a particular case of HITS Kleinberg et al 1999 one of the most famous link analysis algorithmsThe rest of this paper is organized as follows Section 2 discusses papers related with our work In Sect 3 we introduce some basic concepts useful to understand our analytical methodology Section 4 illustrates the process of bipartite network Driver–Place construction while Sect 5 explains in detail the idea of mobility complexity In Sects 6 and 7 we present the experimental results obtained in the two case studies using reallife GPS and GSM data Finally Sect 8 contains conclusions and describes future works
Keywords:
.
|
Other Papers In This Journal:
|