Semantic Trajectories in Mobile Workforce Management Applications

As a consequence of the competition between different manufacturers, current smartphones improve their features continuously and they currently include many sensors. Mobile Workforce Management (MWM) is an industrial process that would benefit highly from the information captured by the sensors of mobile devices. However, there are some problems that prevent MWM software from using this information: i) the abstraction level of the activities currently identified is too low (e.g., moving instead of performing an inspection on a client, or stopped instead of loading a truck in the facility of a client ; ii) research work focuses on using geographic information algorithms on GPS data, or machine learning algorithms on sensor data, but there is little research on combining both types of data; and iii) context information extracted from geographic information providers or MWM software is rarely used. In this paper, we present a new methodology to turn raw data collected from the sensors of mobile devices into trajectories annotated with semantic activities of a high level of abstraction. The methodology is based on activity taxonomies that can be adapted easily to the needs of any company. The activity taxonomies describe the expected values for each of the variables that are collected in the system using predicates defined in a pattern specification language. We also present the functional architecture of a module that combines context information retrieved from MWM software and geographic information providers with data from the sensors of the mobile device of the worker to annotate their trajectories and that can be easily integrated in MWM systems and in the workflow of any company.