Geolocation of Historical Texts Using AI

  Suggested by: Johannes Scholz in cooperation with  the Österreichischer Alpenverein

Source: Österreichischer Alpenverein

Keywords: geoparsing, artificial intelligence, geographic information extraction

Objective: Utilize AI-supported methods to geolocate historical texts, enabling geographic searches that connect archival publications to specific locations for enhanced accessibility.

Short Description: 

Founded in 1862, the Austrian Alpine Club holds countless publications, texts, and technical articles awaiting revitalization in its archives. Narratives and descriptions from our alpine homeland could be made much more accessible if places could tell stories. AI-supported geolocation of these contributions would enable geographic searches on a map, allowing users to find all publications related to a specific location.

Challenges include historical names, varying spellings, and different languages. The results would be applied in an information portal, bringing historical and current texts closer to people by greatly simplifying the search process. A practical application of the research results is certain in this context.

Start: Anytime

Semi-Automated Delineation of Land Cover for Cartographic Use

 Suggested by: Johannes Scholz in cooperation with Dirk Tiede, Martin Sudmanns and the Österreichischer Alpenverein

Source: Österreichischer Alpenverein

Keywords: land cover delineation, cartographic representation, topographic maps

Objective: Create a reliable method to semi-automatically delineate land cover classes from base data for cartographic representation, reducing manual processing and enhancing map production quality.

Short Description: 

Geographic data is ubiquitous in the form of maps, though often presented poorly. To produce high-quality topographic maps, an exact delineation of land cover is required alongside many other processing steps. Distinguishing between forest and dwarf pines ("Latschen"), as well as between rocks and scree, still requires extensive manual work, especially when unsatisfied with poorly digitized OSM polygons. Reliably deriving these classes from available base data and preparing them for cartographic representation would add significant value to cartography.

Subsequent steps in map production, particularly the automated depiction of rocks and scree, rely on these delineations. A functional concept would also be welcomed by our partners, ensuring the practical application of the research results.

Start: Anytime

Automated Rock Drawing in the AV Style

 

Suggested by: Johannes Scholz in cooperation with the Österreichischer Alpenverein

Source: Österreichischer Alpenverein

Keywords: Cartography, Artificial Intelligence, Terrain Representation

Objective: Develop a workflow using AI-supported methods to transfer the traditional rock depiction style to current data foundations, enhancing topographic maps for mountain sports with historical cartographic aesthetics.

Short Description: 

The Austrian Alpine Club (Österreichischer Alpenverein) boasts over 150 years of cartographic tradition. Since Franz Keil's first map in 1856, the Alpine Club's cartography has been rich in history. Figures like Rohn, Finsterwalder, Ebster, and Brandstätter have influenced generations of cartographers in terrain representation and plasticity. This project aims to bring these achievements of cartographic history into the present.

Source: Österreichischer Alpenverein

Using AI-supported methods, a workflow will be developed to transfer the traditional rock depiction style to current data foundations. Topographic maps, whether in print or digital applications, would greatly benefit, offering an optimal planning basis for mountain sports. Historical maps, separated print files, and various base data will serve as training data. By analyzing terrain features (land cover and terrain form), a representation for new rock areas will be generated that includes mountaineering-relevant terrain details and ensures geometric accuracy. Achieving a close approximation to the historical cartographic aesthetic is highly desirable.

Source: Österreichischer Alpenverein

Start: Anytime

Covid-19: Effects of epidemic arrival time on the global variability of excess mortality

Suggested by: Christian Neuwirth (Z_GIS – Spatial Simulation)

Source: The Economist

 

Short description:

Covid-19 is the first digitally documented pandemic in history, presenting a unique opportunity to learn how to best deal with similar crises in the future.

The large global variability in cumulative excess mortality indicates that countries were not equally successful in handling covid-19 outbreaks. Empirical investigations showed that predictors like population age, gross domestic product, quality of the healthcare system or availability of vaccination have an impact on national excess mortality. In contrast to those variables, the effect of epidemic arrival times was rarely considered.

Hypothesis: Locations that were hit early on and unprepared were facing more severe outbreaks (longer duration without vaccination, lack of non-pharmaceutical interventions, additional winter outbreak etc.). Accordingly, early arrival is associated with high epidemic prevalence, reproduction, and excess mortality.

Method: 1) Construct a Global Mobility Network (GMN) from worldwide air travel data (Data: Open Sky COVID-19 Flight Dataset and/or Flight Radar API) that reproduces the situation at the time of outbreak (early 2020); 2) Implement a SIR network simulation model as presented by Brockmann & Helbing (2013); 3) Run outbreak simulations (Wuhan as expected source) to get estimated arrival times (ranks); 4) Validate results by comparing modeled arrivals (ranks) with reported arrival times; 5) Correlate modeled arrivals with reported excess mortality.

References/Suggested reading:

• Brockmann, D., Helbing, D., 2013. The hidden geometry of complex, network-driven contagion phenomena. science 342, 1337–1342.

Start: ASAP


Prerequisites/qualifications: 

Interest in spatial statistics and/or spatial simulation, basic knowledge of R or Python


Please contact Christian Neuwirth in case of interest: christian.neuwirth@plus.ac.at

Visualizing Places

Suggested by: Franz-Benjamin Mocnik

Keywords: places, cartography, user testing

Objective: Developing and comparing different cartographic means to display places on a map.

Short description: 

When talking about spatial (and sometimes also non-spatial) entities, we refer to places.  Enschede, the ITC, shopping areas, and our homes are examples of such places for which objective descriptions are often hard to find.  In these cases, the spatial boundaries are usually unclear, and people have different understandings of what constitutes these places.  Independent of this fact, we are able to successfully refer to places in our everyday language, and we use them to conceptualize our environment.  It is, however, still unclear how places can be represented by formal means and visually be conveyed.

The proposed thesis topic seeks to explore opportunities of visually conveying places.  Aalbers (2014) has discussed several examples of maps that were influenced by and have influenced existing geographies.  The maps named by him, however, make only use of simple cartographic means.  Mocnik and Fairbairn (2018) have explored further ways of how to adapt maps in order to convey more idiosynchratic content.  As part of the thesis project, characteristics of places need to be discussed, as well as corresponding visual properties.  This includes the spatial extent of places, emotional attachments, place types, the identity of places, and various types of relations between places.  When representing such a characteristics of a place on the map, particular attention shall be paid to correspondences between the places and their visual counterparts in terms of affordances, structural referencing, and further means to create analogies.  The different ways of representing places shall be compared by user testing, including eyetracking, interviews, and screen logging.

Literature references: 

• MB Aalbers: Do Maps Make Geography? Part 1: Redlining, Planned Shrinkage, and the Places of Decline. ACME: An International E-Journal for Critical Geographies, 13(4), 2014, 525–556
• FB Mocnik and D Fairbairn: Maps Telling Stories? The Cartographic Journal 55(1), 2018, 36–57

Start: anytime

Maps Telling Stories

Suggested by: Franz-Benjamin Mocnik

Keywords: storytelling, cartography, user testing

Objective: Implementing and testing modifications to existing cartographic means with respect to their ability to tell stories.

Short description:

Stories often convey emotions, and they convey a narrative that make us understand how it would be to be in the position of the protagonist.  By providing a description of an idiosynchratic experience, stories are more than a formal representation of shared conceptualizations of what happens.  This is in contrast to cartographic representations, which in many cases aim to provide an ‘objective’ view of our environment.  If a story shall be convey by a map, other media are usually included.  For instance, multimedia maps can convey stories by adding pictures, videos, and audio recordings.

Mocnik and Fairbairn (2018) have explored novel ways of how to adapt maps in order to make them more text-alike in their structure, thus hoping for being able to convey stories in more engaging and idosynchratic ways.  This thesis project aims at implementing and testing the proposed and similar cartographic means.  For implementing the cartographic means, either the rendering of maps would need to be adapted, or the depiction of prerendered map tiles would be extended by additional elements using Leaflet, D3.js, and similar technologies.  Then, a story would be chosen and visualized by utilizing the implemented cartographic means.  Finally, the produced visualization would be compared to a textual one.  By making use of suitable techniques, inlcuding eyetracking, interviews, and audio recordings, both the implemented visualization and the textual representation will be evaluated in terms of how well they are able to convey emotions and idiosynchratic views.

Literature references: 

•  FB Mocnik and D Fairbairn: Maps Telling Stories? The Cartographic Journal 55(1), 2018, 36–57

Start: anytime

Dimension of Geographical Networks

Suggested by: Franz-Benjamin Mocnik

Keywords: spatial networks, dimension, street network, OpenStreetMap

Objective: Developing and implementing algorithms for detecting subnetworks the local network dimension of which is homogeneous.

Short description:

Networks occur virtually everywhere. The World-Wide Web, metabolic networks, and communciation networks are typical examples.  Also in Geography, such networks occur.  Mocnik (2018a) has examined the degree of influence space has on such networks.  He showed, for instance, that transport and road networks, as well as social networks to a lesser degree, are influenced by space.  Further more, the dimension of the space has a strong impact on the topological structure of such a network, which is why a dimension can be assigned to a network.  Besides exposing very similar qualities, street networks in cities are very similar with respect to their dimensionality (Mocnik 2018b).  In order to model such networks, Mocnik (2015a, 2015b) has proposed a spatial network model.

This thesis aims to develop strategies in order to identify subnetworks that expose a similar dimension in every neighbourhood.  That is, a network that is homogeneous and is thus similar to the spatial network model proposed.  Such networks naturally arise as has been discussed at the example of street networks in cities (Mocnik 2018b), and at the example of road networks as multi-layered networks (Mocnik 2018a).  The thesis will pay particular attention to the example of street networks, which can be extracted from the OpenStreetMap dataset, an open and freely available dataset.  As a result of this thesis, it will be tested whether such homogeneous subnetworks have a geographical relevance and can/should be considered to be geographical entities?  Would such an algorithm, for instance, be able to identify cities by their network dimension?

Literature references: 

• FB Mocnik (2018a): The Polynomial Volume Law of Complex Networks in the Context of Local and Global Optimization. Scientific Reports 8(11274), 2018
• FB Mocnik (2018b): Dimension as an Invariant of Street Networks. Proceedings of the 7th International Conference on Complex Networks and Their Applications, 2018, 455–457
• FB Mocnik (2015a): A Scale-Invariant Spatial Graph Model. PhD Thesis. Vienna University of Technology, 2015
• FB Mocnik, AU Frank (2015b): Modelling Spatial Structures. Proceedings of the 12th Conference on Spatial Information Theory (COSIT), 2015, 44–64
  

Start: anytime

Prerequisites / qualifications: This topic requires some previous knowledge in network science and a good understanding of algorithms.