Humanitarian aid is becoming increasingly important as ever more emergencies such as natural disasters, wars, and pandemics threaten entire societies across the globe. During these emergencies, timely geospatial information about the event is crucial to enhance situational awareness for humanitarian aid. However, current monitoring approaches often cannot provide sufficient, timely, high-quality results for disaster management. In the case of refugee movements caused by such emergencies, there is often hardly any information available. Addressing and overcoming these limitations is critical for humanitarian aid and can be addressed with available big data.
In his dissertation, entitled
"Semantic and Geospatial Machine Learning Analysis of Social Media Data for Humanitarian Aid",
Clemens Havas presented a novel methodology for the analysis of georeferenced social media data, specifically Twitter data, to provide additional information for humanitarian aid. The methodology combined state-of-the-art machine learning methods (topic modelling and hot spot analysis) developed in two different fields, namely natural language processing and geographic information science. Its objective was to extract relevant social media posts within a specified region and period and identify places connected with the analysed event.
Conclusion of this thesis was, that social media analysis can be used as a flexible tool for humanitarian aid and help close the information gap until more data sources emerge and supplement existing findings. Additionally, social media outcomes serve as proxies for real-world geographic phenomena, as shown by comparisons of celled results and actual investigated events.
We wish Clemens all the best for his future career!
