Visual Analytics

Geovisual Analytics for Social Media Data

3502.2 - Using Graphs for the Analysis and Visualization of Communities from Location-Based Social Media Content

Monday, July 3
1:50 PM - 2:10 PM
Location: Delaware B

As users have increasingly adopted location-based social network (LBSN) services where they broadcast user-generated content (UGC), there is a need to understand the impact of geography in the formation of the communities and the propagation of the content. Although statistical analysis is often used for discovering patterns and formulating hypotheses about the social interaction, visual analysis can provide better overviews and reveal patterns missed via quantitative measures alone. As LBSN are built on the foundation of users interconnected through relationships, graphs are the visual representation of interconnected nodes. Thus, graph-based representations are by far the most common form for modeling social networks (SN). But there are only few examples how to combine SNs with the geographic space. LBSN visualizations are often limited by simply showing a user's location and not the connection to other users. Such maps hide a number of information included in the data, especially the formation of communities and the social network structure through which the content is propagated.

In order to detect and reveal different communities and their communication behaviour in the geographic space we use a multilevel approach that combines spatial, semantic, social and temporal parameters from the UGC.
In the first step (semantic-temporal analysis) the UGC is filtered by keywords and time in order to obtain a dataset to a certain topic or event for a certain timeframe. Key themes and subthemes are detected by finding co-occurences of indexed words (hashtags) or word pairs and representing them as a graph.
The second step (social-semantic analysis) comprises of the extraction of the relationships among the users, the detection of the communities based on the connectedness of the users and the following semantic analysis of the communities. This enables a better differentiation and understanding of the detected communities with regard to their views and opinions. Also in this step a graph is used to visualize the social relations.
In step three (spatio-social analysis) the location of the users are extracted either from the profiles or from the geocoded content. After the geocoding of toponyms, the locations are verified by a comparative analysis with geocoded content in the user's content history. Different granularities of the geographical reference are then changed to city-level.
In the fourth and last step (spatio-social-sematic visualization) the results of the previous steps are merged for a cartographic representation. The network structure is visualized as a graph where nodes represent users and edges represent a connection between the users. Rings around nodes aggregate users locating in the same city in order to avoid the occlusion of nodes and to focus users to the city. The color of the nodes indicates the community the user belongs to and the size indicates the node degree. The color of the edges represents the quality of the relation (symmetrical/asymmetrical) and the width the amount of shared content. It is also possible to aggregate users from a city to a single node. All edges will be aggregated as well. Then the nodes represent the users of the different communities as a pie chart whereas the size indicates the number of users locating in this city.

For various applications we used Twitter data in order to highlight the advantages of this cartographic visualization. They clearly show the spatial distribution of the different communities as well as regional hotspots. Users and locations who act as information source and regions of higher exchange of information are quickly recognizable.

Thomas Gründemann

research associate
Technische Universität Dresden

1997 - 2004: Study of Cartography at the University of Technology in Dresden, Germany
2000 - 2001: Research visits at the Department of Geography in Columbus, Ohio and at the Canada Centre for Remote Sensing in Ottawa
2005 - 2006: Research associate at the University of Technology in Dresden, Germany, subject 3D-Visualization
2006 - 2014: CTO of mbmSystems GmbH, Dresden, Germany
Since 2015: Research associate and PhD student at the University of Technology in Dresden, Germany, subject visualization of UGC

Presentation(s):

Send Email for Thomas Gründemann

Daria Hollenstein

Research Associate
FHNW University of Applied Sciences and Arts Northwestern Switzerland

Daria Hollenstein is a Research Associate at the Institute of Geomatics Engineering at the FHNW University of Applied Sciences and Arts Northwestern Switzerland in Muttenz (Basel). Currently, she works for the Geo-visualization and Visual Analytics group at the Institute. She has graduated from the University of Basel with a Master’s degree in Archaeology.

Presentation(s):

Send Email for Daria Hollenstein


Assets

3502.2 - Using Graphs for the Analysis and Visualization of Communities from Location-Based Social Media Content



Attendees who have favorited this

Please enter your access key

The asset you are trying to access is locked. Please enter your access key to unlock.

Send Email for Using Graphs for the Analysis and Visualization of Communities from Location-Based Social Media Content