GEO MASH-UP

Current geographic recommendation engines are rather static and only marginally take into account users' geo-preferences and time-dynamic aspects. Our Geo Mash-Up research addresses both limitations, and investigates several techniques and strategies to engage in more dynamic geographic information processing.

A first focus area includes trajectory (and hotspot location) analysis of recreational outdoor activities using mobile sensors. The resulting data are used to construct a geographic user profile, and to generate a personalized mash-up (i.e., a multimedia summary) of the user's activity by using online geo-tagged media related to the hotspot locations. Geographical preferences are then used to recommend relevant routes/points of interest to other users (with similar profiles).

A second focus area includes the time-dynamic aspect of geographic information and how this can be incorporated into the existing geo-information landscape. The concept of points-of-interest (POIs), for example, is being revisited - and a more dynamic prototype is currently under development.

 

DEMO - iMinds The Conference

During iMinds the conference, we will demonstrate the GEO MASH-UP concept and use participants' gpx-data to generate a multimedia mash-up of their activities and estimate their geographic profiles. Furthermore, we will link their profiles to other participants with similar geo-preferences and propose relevant POIs and (alternative) routes. At the conference, we will showcase your geographic profiles and show the people/places you are 'geographically' connected to.

Select GPX file:
Your email address:
Select your picture (optional):

 

General architecture

As shown in the figure below, the GEOMASHUP architecture consists of four building blocks. The sensor data logging, i.e., the first block, is performed using an in-house Android app, which each five seconds logs the GPS coordinates, the vibration data and the smartphone status. Next, the second block extracts the candidate keypoints, e.g. locations of low-activity, and filters out route-specific keypoints (such as traffic lights) using online trajectory information. In order to retrieve the geotagged media which summarizes the user's activity, the following block then queries a set of social media web services with the coordinates of the remaining candidate keypoints. Finally, the retrieved media objects are combined in a personalized route mashup of the user's activity.

 

Publications about our GIS research

Conference publications:

1. Verstockt, S., Slavkovikj, V., De Potter, P., Slowack, J., and Van de Walle, R., "Multi-modal Bike Sensing for Automatic Geo-annotation: Geo-annotation of Road/Terrain Type by Participatory Bike-sensing," 10th International Conference on Signal Processing and Multimedia Applications (SIGMAP 2013) - BEST PAPER AWARD.  
2. Verstockt, S., Slavkovikj, V., Janssens, O., De Potter, P., Slowack, J., and Van de Walle, R., "Web-based Enrichment of Bike Sensor Data for Automatic Geo-annotation," 2nd ACM SIGSPATIAL International Workshop on Crowdsourced and Volunteered Geographic Information (GEOCROWD 2013) - BEST PAPER AWARD.  
3. Verstockt, S., Slavkovikj, V., De Potter, P., Vandersmissen, B., Slowack, J., and Van de Walle, R., "Automatic GEO-MASHUP generation of outdoor activities," 11th International Conference on Advances in Mobile Computing & Multimedia (MoMM2013).  

Journal publications:

4. Verstockt, S., Slavkovikj, V., De Potter, P., Slowack, J., and Van de Walle, R., "Collaborative Bike Sensing for Automatic Geographic Enrichment," Special Issue on Signal Processing for Big Data - IEEE Signal Processing Magazine (submitted).

Presentations:

4. Verstockt, S., "GEOCROWD PANEL on Social Media Mining and Geospatial Data," 2nd ACM SIGSPATIAL International Workshop on Crowdsourced and Volunteered Geographic Information (GEOCROWD 2013).  

More publications by Verstockt, S. et al. can be found in the UGent database.

 

 

Contact:

dr. Steven Verstockt

Ghent University
Faculty of Engineering
Department of Electronics and Information Systems
Multimedia Lab
Gaston Crommenlaan 8, B-9050 Ledeberg-Ghent, Belgium

p: +32 9 33 14985
f: +32 9 33 14896
e: steven.verstockt@ugent.be