Department of Geography

Geographic Information Systems

Topics for MSc Theses, GIS Unit

March 2017

General Overview: Research in the GIS Unit

Our group develops and utilizes methods that seek structure in spatio-temporal data, thus turning raw
data into geographic information, ultimately aiming at generating knowledge that leads to a better
understanding of geographic patterns and processes. Our research focuses on the following
application areas:
• Computational Movement Analysis and Simulation
• Language and Space
• Location Based Services (LBS) & Computational Cartography

Our methodological toolset draws from an interdisciplinary range of fields, including spatial analysis,
spatial statistics, algorithms development, and computational techniques such as data mining and
agent-based modeling.

Choosing a Topic

We recommend that you first take a look at the list of ongoing and past MSc projects, with
downloads of MSc theses: http://www.geo.uzh.ch/en/units/gis/research/MSc-Theses0.html.

Think about what interests you most, and what you are good at. Do you want to program, are you
good at it, or rather not? Do you like to work empirically, running experiments and analysing the
results, or would you rather develop something new (in which case you would probably have to
program)? Is real-world applicability important to you, or are you ready for blue skies research? Do
you have your own topic, or a topic that you wanted to pursue with a third party (e.g. WSL, ETH)?

Finally, come talk to us. In the topic descriptions below, we have listed the contact persons in our
group. Additionally, Robert Weibel can give an overview of the project topics in the GIS Unit.

Don’t forget that the MSc project is primarily a scientific project. Even if you might be more interested
in applied work, the outcome must be more than what would typically be achieved in GIS projects
undertaken by an engineering firm. Hence, what are the research questions you want to investigate?
We are there to help you formulate suitable research questions and bring your research to fruition.

Seite 1/12
 Department of Geography
Geographic Information Systems

Computational Movement Analysis and Simulation

Trade-offs between precision and aggregation in computational analysis of

human movement
Short description: Human movement is analysed with a wide range of differing methodologies but
usually based on GPS and/or Accelerometer data. Almost always this is done way below the
sampling rates that today’s sensors could easily provide (such as 50Hz). This is mostly to save disk
space and/or bandwidth. For some applications this aggregation is clearly sufficient. Other
applications that have characteristic movements in the sub-second range would potentially benefit
from the higher resolution. One such example would be transport mode detection, where long time
series are important and therefore a high compression is desirable, but it seems plausible that small
but characteristic movement patterns are present and it is unclear whether they can be captured fully
by generic summary statistics.

The aim of this research would be to set up a system for collecting and labelling data of high
granularity, conducting a data collection and labelling campaign and exploring the trade-offs between
data compression and quality of the outcomes in e.g. transport mode detection. Assuming the trade-
off is non-trivial (i.e. the added granularity indeed leads to an improvement of the classification),
implications for future data collection efforts (classification at collection time, additional features to be
collected…) should be thought about and ideally demonstrated.

Methods, requirements: This project requires a sound understanding of machine learning

techniques (and at least one language that lets you use them, such as R, Matlab or Python) as well
as the motivation to conduct a collection campaign.

Language: Thesis can be written in German or English.

Supervisor(s): Robert Weibel, Oliver Burkhard

Initial readings:
Laube, P., Purves, R. (2011): How fast is a cow? Cross-Scale Analysis of Movement Data.
Transactions in GIS. 15(3): 401-418.
Prelipcean, A., Gidófalvi, G., Susilo, Y. (2016): Transportation mode detection – an in-depth review of
applicability and reliabilty. Transport Reviews. http://dx.doi.org/10.1080/01441647.2016.1246489
Hemminki, S., Nurmi, P., Tarkoma S. (2013): Accelerometer-Based Transportation Mode Detection
th
on Smartphones. Proceedings of the 11 ACM Conference on Embedded Networked Sensor
Systems.

Cross-scale analysis and classification of movement

Short description: This topic is partly related to the previous topic (“Trade-offs …”). Movement
tracking data is being generated at increasingly finer time intervals. Depending on the tracking
technology used and the conditions of deployment, sampling rates in the sub-second range are not
uncommon. For instance, video tracking typically works at 30 frames / second, that is, 30 Hz.
Similarly, if a GPS tracker can be recharged daily and the data downloaded at frequent intervals, it is
easy to set the sampling rate to 1 Hz or higher. That is, we find ourselves in a situation where the
movement patterns that we are trying to detect are massively oversampled. As movement patterns
(and the behaviors they represent) most often take place at different temporal scales, oversampling
now offers the possibility to adjust the analysis scale to the adequate temporal scale, either by

Seite 2/12
 Department of Geography
Geographic Information Systems

resampling to coarser resolutions or by using inherently multi-scale methods (e.g. wavelets).

In a recently completed PhD thesis (Soleymani 2016, Soleymani et al. 2017), several multi-scale
methods for movement classification and for movement segmentation have been demonstrated to
outperform existing single-scale methods. An MSc project could take two orientations. It could either
take the new multi-scale methods and their implementations and apply these to different movement
classification problems, further validating and potentially improving these techniques. Or it could
develop these methods further. For instance, the segmentation method (Soleymani et al. 2017) has
potential for further development; multi-scale classification could be linked to geographical context
data; or further sensors (e.g. accelerometer) could be included.

Methods, requirements: This project requires a sound understanding of machine learning

techniques (particularly if the second orientation is chosen), and at least one language that lets you
use them, such as R, Matlab or Python.

Language: Thesis can be written in German or English.

Supervisor(s): Robert Weibel, Oliver Burkhard

Initial readings:
Soleymani, A. (2016): Cross-scale analysis in classification and segmentation of movement. PhD
Thesis, Department of Geography, University of Zurich ! http://www.geo.uzh.ch/dam/jcr:6c19f44d-
1c92-4a02-8345-0ebce51f4f6e/SoleymaniAli_PhDThesis_Final.pdf
Soleymani, A., Pennekamp, F., Dodge, S. & Weibel, R. (2017): Characterizing change points and
continuous transitions in movement behaviors using wavelet decomposition. Methods in Ecology and
Evolution. DOI: 10.1111/2041-210X.12755.

Mobility, Activity and Social Interaction Study of healthy older adults (MOASIS)
Short description: MOASIS collects individualized everyday-life health data in older adults. The
project started in August 2015 as a collaboration between researchers from GIS and the
Gerontopsychology Group at the Department of Psychology UZH. It ultimately aims to develop
computational models to measure, analyse, and improve health behaviors and health outcomes in the
everyday life of aging individuals. The study design of MOASIS includes baseline tests, self-reports,
and an evening questionnaire, complemented by the ambulatory assessment of the physical
(accelerometer), spatial (GPS) and social activity (audio) with the custom-built sensor uTrail.

Within the framework of MOASIS, potential research topics for MSc projects could be:

• Mapping places of social interaction. Approximately 4 times an hour, 30 second sound

snippets of the participants’ environment are captured. The audio data contains a lot of
meaningful information and analysis of spoken content gives psychologists interesting
insights into people’s everyday activities. These analyses, however, are all based on manual
transcriptions. The idea of this MSc topic would be to find ways to automatically derive
information of these audio files (e.g., based on noise levels) that give indication on potential
social interaction. In a second step the audio data can be linked to the simultaneously
assessed GPS data and places of different types of social interaction could be derived
thereof.
• Transportation mode / physical activity level / type detection based on uTrail data (GPS/ACC):
In MOASIS, we have no ground truth data regarding modes of transport / types / intensity of
physical activity. However, being able to reliably detect the participants’ activities based on
GPS and ACC data is of major interest. In the framework of a MSc project, data consisting of

Seite 3/12
 Department of Geography
Geographic Information Systems

a labeled set of different activities tracked with the uTrail could be gathered in a data
collection campaign administered by the MSc student. Based on the GPS/ACC data
assessed, different methods of activity detection could be applied and validated based on the
ground truth labels. Note that this topic is related to the above topic “Trade-offs between
precision and aggregation …”, but takes a somewhat more pragmatic approach and is
focused on the particular needs of the MOASIS project.
• In the health sciences, accelerometers have been used extensively to measure the level of
physical activity (PA). In GIScience, on the other hand, accelerometry data has long gone
largely unnoticed; unlike GPS data, it lacks the ‘geographical’ scale. However, in order to
detect movement behaviors at the micro level, such as fine-grained PA types (e.g. different
modes of locomotion, or activities of daily living), accelerometer data is indispensable. It also
offers new dimensions for geographers, for instance by detecting PA types (possibly involving
also GPS) and by relating these to geographical context (i.e. which PA types takes place in
which environments? in which places? under which conditions?).
• More topics can be designed in coordination with the ongoing PhD projects of Hoda
Allahbakhshi and Michelle Fillekes. Since MOASIS is a long-term project, lasting at least until
the end of 2019, more topics and research problems are expected to evolve continuously

Methods, requirements: Depending on the focus, a combination of spatial statistics and empirical
analysis in R or Matlab. Potentially machine learning methods for classification of either audio or
movement and accelerometer data are included. A genuine interest in working in an interdisciplinary
setting is a prerequisite. You are not afraid of statistics and getting your hands “dirty” with some
programming. Some basic reading in the field of movement and potentially psychology will be
necessary.

Language: A good command of English is a prerequisite.

Supervisor(s): Robert Weibel, Hoda Allahbakhshi, and/or Michelle Fillekes (depending on the topic)

Assessment of older patients’ real-life mobility by the general practitioner:

Making use of modern technology
Short description: So far, the use of GPS-derived movement parameters to quantify physical
performance has mainly been limited to team sports. In the health sciences, most applications have
been confined to estimating activity spaces of individuals from GPS fixes, often linking these to active
transport and body weight. There are only very few reports on applications of GPS-derived movement
parameters in patient populations. So far, GPS-derived movement parameters have not been tested
for validity and for how results compare to traditional measures of mobility function used in the health
sciences.

While deriving speed and other movement parameters from consumer-level GPS (e.g. in
smartphones or mid-range trackers) – due to its accuracy in the meter range – is feasible only over
longer distances covered and at higher speeds, locomotion speed can be accurately extracted from
accelerometer readings already over shorter distances. Since 3-axial accelerometers, like GPS, today
are a standard component of contemporary smartphones, there is a potential for ACC measurements
to be used to replace or complement traditional walking tests over short distances (4, 10, 20 m) used
in the health sciences. Since accelerometers do not rely on an external referencing system – in
contrast to GPS devices which require visibility of GPS satellites – they can be used both indoors and
outdoors. On the other hand, GPS is the optimal method to determine locations over longer periods of
time, and therefore the optimal method to assess life-space mobility.

Seite 4/12
 Department of Geography
Geographic Information Systems

Methods, requirements: This project would first implement a method to derive speed from ACC
measurements and then experimentally assess the validity and reliability of the results compared to
traditional walking tests. Furthermore, it would link these ACC-derived speeds to GPS measurements
in order to develop a smartphone-based mobility assessment methodology that has the potential to
replace existing traditional approaches, thus finding a more wide-spread deployment among general
practitioners.

Language: Thesis should be written in English.

Supervisor(s): Robert Weibel, Timo Hinrichs (Department of Sport, Exercise and Health, University
of Basel)

Initial readings:

Bertschi M et al. (2015). Accurate walking and running speed estimation using wrist inertial data. 3rd
Annual IEEE Intl. Conference on Engineering in Medicine and Biology Society (EMBC), 8083-6.

Wilson AM et al. (2013). Locomotion dynamics of hunting in wild cheetahs. Nature 498(7453): 185-9.

Do Taxi Drivers Take the Fastest Routes? – A Large-Scale Analysis Using

GPS-based Floating Car Data in Vienna
Short description: Understanding the nature of taxi drivers’ route choice behavior is essential for
traffic modeling as well as the development of intelligent transportation systems. On the other hand,
studying how taxi drivers make route decisions will also provide important insights to improve existing
car navigation systems, which so far mostly provide shortest or fastest routes.

This project aims to analyze how taxi drivers make route choice decisions when they have
passengers onboard, using a large-scale FCD (floating car data) dataset in Vienna (Austria).
Particularly, the following research questions will be addressed: Do taxi drivers with passengers
onboard take the fastest routes? How do the actually chosen routes differ from their corresponding
fastest routes? What are the route characteristics preferred by taxi drivers with passengers onboard?

Methods, requirements: This project will focus on computational movement analysis, especially on
big data analytics. Programming skills are required, at least in a scripting language.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang, Robert Weibel

Modelling Urban Semantics and Mobility from Heterogeneous Crowdsensed

Data
Short description: Thanks to the massive crowdsensed data collected in urban spaces, we can now
understand human mobility patterns, urban dynamics, and spatial interactions from a new perspective.
Existing research on this aspect has mainly focused on using a single type of data, particularly either
on GPS data or social media data, but not both.

This research aims to integrate heterogeneous data sources to provide a more comprehensive
picture of how people behave in the urban environment, particularly on urban semantics (e.g. how is
the city used by its inhabitants, and does this relate to urban functional zones?) and urban mobility

Seite 5/12
 Department of Geography
Geographic Information Systems

(how do people move around the city?). The following types of crowdsensed data will be used and
integrated: call detail record (CDR) data (i.e. mobile phone data), passenger flow data of public
transportation (via smart IC cards), taxi GPS data (floating car data, FCD) and bus GPS data. A
tentative study area will be the city of Shenzhen (China), which is one of the five largest cities in
China, and is located immediately north to the Hong Kong SAR.

Methods, requirements: This project will focus on spatio-temporal data analysis and computational
movement analysis, especially on big data analytics. Programming skills are required, at least in a
scripting language.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang, Robert Weibel

Behavioral classification of animal movement data

Short description: Tracking data recording the trajectories of animal movement are becoming
increasingly available nowadays, together with data collected from other sensors (accelerometer, etc).
In different domains of movement ecology, the interest is on the extraction of behaviors from such
trajectories and possibly linking them to the environmental factors or to the patterns extracted from
other sensors. Although the data might refer to different species, the aims of behavioral classification
remain largely the same: developing methods that are capable of detecting relevant behaviors in
trajectories of animal movement. This can be accomplished through segmentation of trajectories into
different sections, analysis of movement parameters (speed, acceleration, turning angle, etc.), and
the use of machine learning algorithms for the classification.

Potential projects: We are currently collaborating with several groups of animal ecologists, who
have collected data and who are interested in getting help in spatio-temporal data analysis. Examples
of past and ongoing MSc projects can be found at http://www.geo.uzh.ch/en/units/gis/research/MSc-
Theses0.html. The precise topic of a new MSc project will be defined in collaboration with the external
animal ecology group.

Methods, requirements: Computational movement analysis; machine learning (e.g. using

RapidMiner, R, Matlab); statistical analysis (using R); programming in R and/or Matlab (or Python or
Java)

Language: A good command of English is a prerequisite, as you will be collaborating with

international groups.

Supervisor(s): Robert Weibel, and the corresponding (external) animal ecology expert

Additional remarks: A visit to the field site(s) of the species under study is an option.

Seite 6/12
 Department of Geography
Geographic Information Systems

Language and Space

“Language and Space” broadly describes the interdisciplinary field at the intersection between the
spatial sciences and linguistics. MSc projects in “Language and Space” can either have a primary
focus on the spatial sciences, with language being used as input information or, the focus lies on
research questions deriving from linguistics and the goal of the MSc project is to fruitfully apply
methods from the spatial sciences to linguistic data. The two types of projects are truly
interdisciplinary in the sense that the master student will be supervised by researchers from different
disciplines. This is ensured by the fact that our group participates in the University Research Priority
Program “Language and Space” of UZH and currently also pursues two Swiss National Science
Foundation projects jointly with research groups in linguistics. In the following, both types of projects
will briefly be introduced and exemplified. Precise project definitions would be developed in
discussion with the supervisors named below.

Spatial analysis and linguistic hypotheses

Linguistics is in the favorable situation that rich information on dialects and languages that has been
collected in laborious field-work over the last century, has only recently been summarized and made
available in large data bases. This information offers the opportunity to quantitatively test the wealth
of linguistic hypotheses on how language evolved over space and time. In the following some
hypotheses that might be tested in an MSc thesis are described:
• Accessibility and linguistic complexity: Remoteness, rough terrain or deep vegetation, in short:
difficult accessibility, have often been argued to be key drivers of linguistic complexity.
Languages in regions difficult to reach are said to have retained a richer morphology
compared to languages in easily accessible regions. Potential case study regions for testing
this hypothesis are South East Asia (incl. the Himalayas), South America or Italy (dialects).
• Geography, a linguistic border: The change from one dialect to another is often smooth and
continuous. Also, adjacent languages tend to share many grammatical or lexical
characteristics. At times, however, we observe sudden abrupt changes, be it in dialects or
adjacent languages. Such abrupt language changes are often associated with geographical
(e.g. mountain ranges, large rivers), administrative (states or cantons) or cultural (e.g. religion)
borders. There is a wealth of linguistic data available that allows to test this hypothesis in
different regions of the world (incl. Switzerland) and at different spatial scales.
• Language Change and Mobility: For a long time, language has been projected to space
without further consideration of speakers. Newer theories, however, suggest that
incorporating the speaker and his/her background might offer new insights into so far poorly
understood patterns in the linguistic distribution. It is for instance assumed that local mobility
behavior (such as excessive commuting, migration, etc.) has a strong impact on how — and
in which direction — dialects or languages change. We have detailed linguistic information for
the whole of German-speaking Europe that covers the last 100 years. Linking this data to
local sociodemographic variables might prove to be a fruitful means for discovering new
systematics in language change.
• Sprachbünde: There is a body of literature and theories on similarities of languages on a
global scale (so-called Sprachbünde). However, most empirical work in linguistics exclusively
focuses on single language families (hundreds of language families exist, sometimes
counting up to 1000 languages). The aim of this strand of MSc projects would be to conduct
quantitative studies on a global scale and thus test the hypothesis of larger Sprachbünde, for
instance in Eurasia or the Transpacific Belt.
• Language diffusion and contact in space and time
Understanding the diffusion and contact of languages is essential to understanding the
diffusion of humanity. Human migration creates the potential for language contact and

Seite 7/12
 Department of Geography
Geographic Information Systems

facilitates the propagation of linguistic features. When we observe evidence for language
diffusion and language contact, we observe evidence for human migration. Human migration
is influenced by both the environment (e.g. topography, land cover) as well as cultural factors
(e.g. population pressure, the emergence of empires). However, the role of geography for
human migration is not always clear. The aim of this MSc topic is to test hypotheses of
human migration in space and time using evidence from language diffusion and contact.
Which spatial factors have facilitated or constrained human migration? For example, do
linguistic features propagate along river networks, such as the Amazon?

This list of topics is not exhaustive and only gives a first overview of potential focus areas. All topics
are motivated by a broad linguistic hypothesis, followed by a detailed and extensive spatial analysis.
Examples of past MSc projects: http://www.geo.uzh.ch/en/units/gis/research/MSc-Theses0.html.

Methods, requirements: A basic interest in languages. A flair for spatial analysis and spatial
statistics. Little to moderate programming skills. Experience in working with large data is helpful.

Supervisors: Curdin Derungs, Peter Ranacher or Robert Weibel, plus a co-supervisor from
Dialectology or Linguistic Typology.

Language as a source of information

Language is the most commonly used tool for communication available to humans. Therefore,
information gained from language use might represent how we make sense of the physical and social
world. The usefulness of such information in numerous applications has long been recognized and is
often associated with the label “data sciences”.
In the spatial sciences we are most interested in how the world — read space — is perceived and
conceptualized in language and how such information can be automatically retrieved. MSc projects
might focus on one of the following issues:
• How does the description and perception of landscapes change over time? Can such change
be automatically studied, for instance from historical books? Books (nowadays in digital
format) can have considerably long timelines and thus allow to look far beyond the temporal
coverage of state of the art sensors in geography. The aim of this strand of MSc projects is to
use historical, digitized text as a means for automatic spatio-temporal analysis of social (e.g.
historical jurisdiction) or physical (e.g. historical alpine guide books) processes.
• What is the geographic meaning of word frequencies? Word frequencies are an established
measure in the data sciences for gaining an insight into the meaning of text. However, studies
on the relation between word frequencies and for instance characteristics of landscapes that
are described in text have not gained much attention.
• Can the meaning of spatial relations, such as ‘near’, ‘north’ or ‘besides’, be investigated from
large language data, such as web pages, digitized books or social media? Spatial relations
are particularly prone to vary due to individual interpretation (What is ‘near’ and in what
context?). However, understanding spatial relations is crucial for automatically retrieving
spatial information from text. Additionally, text offers an under-investigated but very rich
source of information on how people use spatial relations in different situations.
• How do people describe familiar places, such as public spaces in a city, and how do these
descriptions reflect the characteristics and aspects of space perceived as being most crucial?

All of these examples are quite general and need further specification. If additional information is
desired please don’t hesitate to contact the supervisors listed below.

Methods, requirements: Skills in programming. First experience in natural language processing is

helpful but not a requirement.

Supervisors: Curdin Derungs, Ross Purves or Robert Weibel, plus a co-supervisor from
Computational or Corpus Linguistics.

Seite 8/12
 Department of Geography
Geographic Information Systems

Location-based Services (LBS) & Computational Cartography

Indoor Route Planning

Short description: In daily life, people often encounter navigation problems when arriving at a new
place, e.g. “What’s the way from the train station to the city hall?”. Outdoor navigation has been a
research focus since the advent of the first location-based services (LBS). However, after arriving at a
destination by using outdoor navigation services, people need to enter a building and start indoor
navigation. Research has found that compared to outdoors, people tend to lose orientation a lot
easier within buildings, especially complex ones, such as airport, hospitals, and university buildings.
Indoor navigation systems are specifically designed to assist people’s wayfinding tasks in such indoor
environments. However, current indoor navigation systems only provide users with shortest or fastest
routes. Yet, humans rarely move around using only these criteria.

This research will explore methods to provide routes with other characteristics (e.g., simplicity,
fewest-turns and floor-first, and main-corridor-first), based on existing indoor data models (e.g.,
IndoorGML). The Irchel Campus of UZH (or other similar public places) will be used as a test area.

Methods, requirements: Starting from analysis of existing indoor data models, this research will
computationally formulate route choice criteria, and develop route-planning algorithms to provide
routes other than the shortest route.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang, Robert Weibel.

Initial readings:
Golledge, R. G. (1995). Path selection and route preference in human navigation: A progress
report. In A. Frank & W. Kuhn (Eds.), Spatial Information Theory A Theoretical Basis for GIS
(Vol. 988, pp. 207-222). Heidelberg: Springer.

How do People Communicate Indoor Route Instructions?

Short description: Currently, semantics-enriched navigation systems become more and more
popular. Instead of providing metric-based route instructions such as "walk straight in 100 meters",
semantics-based navigation systems provide users with semantically enriched instructions, such as
"walk straight, pass the theatre, and walk to the crossing" or simply "cross the park". This kind of
guidance is more natural to users, and can improve users' navigation performance.

This research aims at empirically answering some fundamental questions of indoor semantic
wayfinding, such as “What are the direction and motion concepts used by humans to communicate
route instructions in indoor environments?” and “What are indoor landmarks?”. Specifically, user
studies will be implemented to collect participants’ route instructions in-situ. These collected
instructions will then be used to derive fundamental linguistic elements: motion concepts (verb, e.g.,
walk, cross, turn, pass), direction concepts (prepositions/adverbs, e.g., straight, along, through), and
landmark concepts (nouns, e.g., WC, lobby, printer). The instructions will be also analyzed to see how
these different kinds of concepts can be meaningfully combined for indoor route guidance.

Seite 9/12
 Department of Geography
Geographic Information Systems

Methods, requirements: This project will focus on user experiments. Methodology and approaches
from a previous research project SemWay (mainly focusing on outdoors; refer to the papers below for
more details) will be reused for the indoor environment.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang, Sara I. Fabrikant

Initial readings:
Karl Rehrl, Sven Leitinger, Georg Gartner, Felix Ortag (2009): An analysis of direction and motion
concepts in verbal descriptions of route choices. In: K. Stewart Hornsby et al. (Eds.): COSIT 2009,
LNCS 5756, pp. 471-488, Springer-Verlag, Berlin-Heidelberg.
Gartner, G., Huang, H., Millonig, A., Schmidt, M. & Ortag, F. (2011): Human-centred mobile
pedestrian navigation system. Mitteilungen der Österreichischen Geographischen Gesellschaft
(Communications of Austrian Geographical Society), 153.

Mobile Maps for Supporting Mixed Indoor/Outdoor Navigation

Short description: Mobile navigation systems are designed to facilitate users’ wayfinding tasks in
unfamiliar environments. Current navigation systems either deal exclusively with indoor or outdoor
navigation. An integration of both versions into one single system has hardly been considered.
However, daily wayfinding often involves mixed outdoor/indoor environments. Obviously, both types
of environments often have different presentation forms. The scale of an indoor map is naturally
larger than the scale of an outdoor map, and therefore a seamless switch between an indoor route
map and outdoor route map is a challenging task. It is still unclear how this switch could be realized
without an abrupt visual switch that potentially leaves the user disoriented. This research will also
explore techniques to support a seamless visual transition between outdoor and indoor navigation.
The proposed technique will be then evaluated in a user experiment.

Methods, requirements: This project will involve mobile map design as well as user experiments.
You should therefore have an interest in empirical cognitive studies as well as the methods of user
testing.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang, Sara I. Fabrikant

Characterizing the Service Interface of Indoor Positioning Methods

Short description: Location-based services have become more and more popular not only in
citywide outdoor environments, but also in shopping malls, museums, and many other indoor
environments. Compared to outdoor LBS, which often employ GPS to obtain users’ current location,
indoor LBS need to rely on other technologies, such as WiFi, Bluetooth and radio-frequency
identification (RFID). Currently, different indoor positioning solutions have been proposed, while a
universal solution such as GPS is still missing. This effectively hinders the current development of
indoor LBS, as users will always need to switch the indoor positioning methods when entering a new
indoor environment. If we can “standardize” the service interface of different indoor positioning
solutions (such as GPS for outdoor environments, or OGC’s WMS, WFS, WCS and WPS, which

Seite 10/12
 Department of Geography
Geographic Information Systems

standardize the input/output of geospatial data and processing services), then an indoor LBS
application would work smoothly across different indoor environments with different positioning
solutions.

As a step further towards this ‘standardization’, the research aims to study what metadata are needed
to characterize and specify the service interface of different indoor positioning solutions. Existing
standards on geospatial metadata will be adapted and extended.

Methods, requirements: This research will involve analysis of diverse indoor LBS applications and
indoor positioning methods, as well as adaptation of existing geospatial metadata standards.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang

Initial readings:
Geospatial metadata, https://en.wikipedia.org/wiki/Geospatial_metadata

MSc Projects in Collaboration with Parkbob

Short description: Parkbob (http://www.parkbob.com/) is a fast growing LBS company, focusing on
providing context-aware on-street parking services. Based on crowdsensing approaches, Parkbob
permanently collects parking activity data from smartphone sensors of its 250,000 monthly active
users. Together with other data sources, these sensors are then used to derive parking events and
parking regulations/rules of particular spots, and finally provide real-time parking services.

Methods, requirements: Depending on the topic, different methods will be used and different skills
required. Mostly, the topics will be related to crowdsensing, mobile sensor data fusion and analysis
(e.g., detection of parking events and parking regulations/rules), and context-aware prediction of real-
time parking availability.

Language: Thesis should be written in English.

Supervisor(s): Haosheng Huang and Robert Weibel, with support from the corresponding contact
persons at Parkbob.

User Testing of Algorithms for Real-time Generalization in Mobile Mapping

Short description: A PhD project has recently been completed in our group by Pia Bereuter, which
resulted in a number of novel algorithms for real-time generalization of point data, such as the point
data that are frequently used as foreground in map mashups of Google Maps, OSM and other online
map services. In order to achieve real-time performance, some simplifying assumptions were made in
comparison to traditional cartographic generalization, and alternative approaches, such as space
deformation (focus+context methods, Bereuter and Weibel 2017), or content zooming (Bereuter et al.
2014), have been introduced.

The questions is now, how effective are these methods in providing information to the mobile user
and supporting his/her information seeking process? How effectively is the user supported in
particular tasks such as wayfinding? How accurately can the user estimate distances, directions and
quantities? In this project, user experiments will be employed to establish evidence about these or
similar questions.

Seite 11/12
 Department of Geography
Geographic Information Systems

Methods, requirements: This project will focus on user experiments. You should therefore have an
interest in empirical cognitive studies as well as the methods of user testing. Preferably, you have
taken GEO 878 (Geovis) and GEO 884 (LBS).

Language: Thesis can be written in German or English.

Supervisor(s): Robert Weibel and Sara Fabrikant

Initial readings:
Bereuter, P. & Weibel, R. (2013). Real-time generalization of point data in mobile and web mapping
using quadtrees. Cartography and Geographic Information Science, 40(4): 271-281.
Bereuter, P., Weibel, R. & Burghardt, D. (2014). Content zooming and information exploration for
mobile maps. International Journal of Geomatics and Spatial Analysis / Revue internationale de
géomatique. 23(3-4).
Bereuter, P. & Weibel, R. (2017). Variable-scale maps in real-time generalisation using a quadtree
data structure and space deforming algorithms. International Journal of Cartography, 3(1).

MSc Projects in Collaboration with swisstopo (swisstopoEDU)

Short description: The Swiss national mapping agency swisstopo has a program for collaboration in
the framework of Masters projects, called swisstopoEDU. In 2013, two MSc students from the GIS
Unit at GIUZ have won a swisstopoEDU award with their respective projects. It would be great to
repeat that experience with new projects.

Methods, requirements: Depending on the topic, different methods will be used and different skills
required. For projects linked to the GIS Unit, the focus will mainly be on topics of digital cartography
(automated generalization and symbolization of TLM data, mobile cartography, geodata in education).

Language: Thesis can be written in German or English.

Supervisor(s): Robert Weibel, with support from the corresponding contact person at swisstopo. Or
other staff member from GIS, GIVA or Gecomputation, depending on the topic chosen.

Additional remarks:
Information about the swisstopoEDU program, as well as a description of currently available topics,
can be found at http://www.swisstopo.admin.ch/internet/swisstopo/de/home/topics/stedu.html

Seite 12/12