Swiss Competence Center for Energy Research
Efficient Technologies and Systems for Mobility
Fig. 1: A Framework for Mobility User Profiling with
Users of Location-Based Services (LBS) increasingly expect them to be
personalized in the sense of being tailored to their individual needs and
preferences. As a prerequisite for successful personalization, the
process of user profiling extracts and stores a set of rules, settings,
needs, interests, behaviors and preferences which virtually represent
each user, and are often based on monitored user behavior (Cufoglu
2012). These issues are also of relevance for the GoEco! project, which
aims at providing users with more energy-efficient travel alternatives
via a mobile app (Cellina 2016). Before this background, we propose to
derive users’ mobility profiles on the basis of their movement
trajectories. As a first step towards that goal, a general concept is
outlined and specific research challenges are identified.
Location-based User Profiling for Personalized Mobility Support
There is growing recognition of our mobility behavior to
differ to a great degree, with individual needs,
preferences or restrictions influencing our choice of
destination, transport mode, time of travel and route
(Golledge and Stimson 1997). Thus, mobility support
systems such as GoEco! need to acknowledge our
heterogeneity for providing better travel
recommendations. A possible information source for
user profiling are prerecorded movement trajectories of
individual users. This poster presents a general concept
for such location-based user profiling (LBUP) and
identifies specific research challenges.
David Jonietz, Dominik Bucher, Martin Raubal
Institute of Cartography and Geoinformation
firstname.lastname@example.org, email@example.com, firstname.lastname@example.org
User Profiling and Personalization
User profiling and personalization involve three phases,
namely information gathering, user profile construction
and finally personalization of services (Gauch et al.
2007). Methods for the former include explicit user
surveying, implicit monitoring of user behavior, or a
hybrid method. Building profiles by mining monitored
data is generally preferable since it is non-intrusive and
allows for dynamic profile updates, however, performs
poorly in case of data scarcity or unpredictable user
behavior (Cufoglu 2014).
Movement Data Analysis
GPS-based movement tracking in the sense of
automatically recording x, y, z coordinate tuples at
predefined time intervals is possible with almost every
modern smart phone. For mobility research, these data
are particularly interesting since they capture human
movement at a very high level of detail. Thus, trajectory
data mining methods have been developed to discover
knowledge from such movement data (Zheng 2016).
There are, however, practical challenges due to data
uncertainty caused by missing data, accuracy problems
or precision deficiency (Andrienko et al. 2016), but also
privacy concerns (Zheng 2016).
There are several research challenges to be addressed
when using trajectory data for user profiling, including
•Data accuracy: The accuracy of the user profile
depends on the accuracy of the (often uncertain and
inaccurate) input data.
•Data sparsity and new user problem: A certain
minimum amount of input data is needed for the
initial creation of a user profile.
•Static vs. dynamic user information: User
information can be more (e.g. modal choice) or less
(e.g. home location) likely to change, also due to
contextual influences (e.g. weather).
•Ad hoc user behavior: Human behavior is not always
optimized and rational, which needs to be
acknowledged in user profiling.
•Privacy: The users’privacy need to be protected.
A General Concept for LBUP Expected Impact
Personalization is highly relevant for mobility support
services such as GoEco!. Travel recommendations which
explicitly acknowledge the specific preferences, needs
and restrictions of the individual user are more likely to
be accepted and lead to a behavioral change towards
more sustainable mobility options. By extracting the
necessary profile information from movement
trajectories, no additional burden is placed on the users
while at the same time, the user profile can be
dynamically updated using a learning approach.
Cufoglu A (2014) User profiling –a short review. International Journal
of Computer Applications 108(3): 1-9
Golledge RG, Stimson RJ (1997) Spatial behavior: A geographic
perspective. The Guildford Press, New York, London
Gauch S, Speretta M, Chandramouli A, Micarelli A (2007) User profiles
for personalized information access. In: The adaptive web,
Springer: Berlin, Heidelberg, 54-89
Zheng Y (2016) Trajectory data mining: An overview. ACM Trans.
Intell. Syst. Technol. 6(3): 29(1)-29(41)
Andrienko G, Andrienko N, Fuchs G (2016) Understanding movement
data quality. Journal of Location Based Services 10(1): 31-46