Mining social media for effective crisis response: machine learning and disaster response

Abstract

Crises in the form of natural or man-made disasters and political upheaval are becoming more common, and events that happen in far corners of the earth are increasingly visible to global audiences. Because of the expected developments associated with climate change, such crises are expected to become more frequent and have greater impacts. Researchers from Trilateral Research undertook an examination of the social impacts associated with the use of social media data to create maps of crisis situations as they develop and their immediate aftermath and the role of machine learning to automatically process data emerging from social media. This research found that although significant potential harms associated with privacy, discrimination and data inaccuracy emerge, the involvement of the humanitarian community alongside authorities and governments has resulted in these issues being both recognised and actively managed.

Authors

Rachel Finn, Hayley Watson, Kush Wadhwa

Date Published

August 30, 2017