Description
Date depot: 1 janvier 1900
Titre: Understanding social and community dynamics from taxi GPS data
Directrice de thèse:
Tulin ATMACA (SAMOVAR)
Domaine scientifique: Sciences et technologies de l'information et de la communication
Thématique CNRS : Non defini
Resumé:
Digital traces are left when people interacting with cyber-physical worlds, which provide us with unprecedented opportunity to better understand our living world. Interpretation of the data is essential. However, most of these raw digital traces have very few semantic meanings, and cannot be used directly. The goal of my thesis is to study algorithms to fill “gaps” between raw digital traces and high-level context, uncovering the hidden facts behind the data itself. What is the high-level context depends on the applications, so the research work in the thesis is application-motivated. But all of them follow the same line of inferring high-level context from digital traces. Two kinds of high-level context have been inferred to support specific application. One work has been done is about identifying meaningful logical location to improve location-based service. In this work, we propose new solution based on digital traces collecting from GPS-enabled mobile phone and wearable camera. We also evaluate the pro- posed solution in various real-life situations, and it is proved to be robust. The other work is about detecting anomalous trajectories from large-scale taxi GPS traces. In this work, we devise an on-line anomaly detector based on isolation mechanism which can not only detect the degree of anomalousness, but also in- form which parts are responsible for anomaly. The proposed approach achieves state-in-art performance through extensive experiments. The future work will focus on working towards the taxi digital traces. And the high-level context can be good money-earning strategies, passenger flow estimation, passenger-finding skills, and so on.
Doctorant.e: Chen Chao