Microsoft developers are aiding in the mission set forth by the United Nations to monitor humanitarian crises around the world in order to establish when and where to send aid.
Microsoft developers were tasked with automating and streamlining the monitoring processes of the United Nations Office for the Coordination of Humanitarian Affairs (OCHA) in Libya to help reduce response time to humanitarian crises using Machine Learning, Open Source Databases, Natural Language Processing, and Microsoft Azure.
Software development engineer Anastasiia Zolochevska explained in Microsoft’s Real Life Code blog the entire process of applying the power of code to potentially save lives.
The early detection system Project Fortis (not to be confused with the Armed Forces veterans community of the same name) identified the UN’s problem of trying to manually monitor over 400 social media and public data sources, and it came up with a more automated solution.
“We built a real-time processing pipeline to monitor factors that contribute to humanitarian crises through conversations and posts from publicly and privately available data sources, including social media,” wrote Zolochevska, adding, “we implemented a system that allows users to define a set of relevant keywords, geographical areas of interest and social media feeds. We then use these parameters to extract related posts and plot them on the map, based on an NLP location feature extraction approach.”
Through a process called “geocoding,” the team was able to utilize a free service that provides worldwide basemap coverage sourced from OpenStreetMap and other open data projects called Mapzen to define an area of interest and its surrounding localities.
Once the geocoding was able to define the region of interest to monitor, the next step was to use the Microsoft Translation Service to be able to translate social media posts and other public sources coming from the targeted region. The Real Life Code article used Libya as an example region with Arabic being the language to translate, hence the need for the translation service.
After carefully aligning the region of interest, translating the public data sources and social media posts, and further pinpointing and cross-referencing the geographical coordinates using an open-source JavaScript library called Compromise, the last step for the team was to use Microsoft Azure Functions as a “hosting mechanism for the locality inference job.”
Monitoring posts across social media platforms using Azure Functions in conjunction with an Azure Event Hub allowed the team to process “a huge volume of social media posts” to help the UN mission in Libya.
The results from this type of project could potentially set a new precedence in how the UN monitors crisis zones by utilizing Machine Learning, open source databases, Natural Learning Processes, and cloud-based hosting platforms like Microsoft Azure.
Through the power of code, Microsoft and the UN are looking to save lives and prevent humanitarian crises.
