Data scientists to the rescue | University of Oxford

Data scientists to the rescue

A unique project involving University of Oxford Information Engineers is integrating the skills of computers and people to enable a more effective response to natural disasters.

Data scientists to the rescueDisasters such as earthquakes or floods pose huge challenges for crisis response teams.
Disasters such as earthquakes or floods pose huge challenges for crisis response teams, who need to know as quickly as possible the extent of the disaster, what aid is required and where they need to get to. The problem is not so much lack of information as information overload. Huge quantities of relevant but unstructured data are quickly generated from photographs, CCTV, news reports, social media, sensors, first responder reports and satellite images. The difficulty is how to process this data deluge in a way that can best inform rescue efforts.

The solution might seem obvious: use computers to do the data analysis. But computers by themselves are not good at finding meaningful patterns in such large amounts of unstructured data, or understanding the complex human problems described within it. In addition, it can be difficult to determine which data and observers are most trustworthy.

Researchers on the ORCHID project (a collaboration between the Department of Engineering Science and Nottingham and Southampton Universities) have developed a truly innovative approach to this problem by fusing the skills of human beings and computers. A ‘crowd’ is recruited through online platforms to analyse unstructured data such as live satellite images, texts and tweets, or detailed video data from drones. Sophisticated machine learning tools employing probabilistic reasoning are then used to reconcile inconsistent responses, filter the reliable sources and aggregate the data. This generates the best information possible from the smallest number of imperfect observers. Sophisticated spatio-temporal models built into the software are even able to interpolate what may lie in the data ‘black spots’ between known observations, thus helping disaster responders decide where they need to gather more information, and where they have a good enough picture.

The approach was trialled in earnest during the first few days of Nepal’s 7.8 magnitude earthquake in April 2015. Working in partnership with Zooniverse, the world’s largest citizen science web portal, and Rescue Global, a respected international crisis response charity, a small ‘crowd’ was quickly recruited to analyse poor-quality satellite images of devastated areas. ORCHID software filtered and aggregated their responses, and blended them with existing data – which identified settlements that did not appear on any maps. Creating this new information enabled responders to be deployed to reconnoitre the settlements and deliver life-saving aid such as water filters.

This new technology offers an evidence-based, rational approach to disaster management. Through collaboration with crisis responders like Rescue Global, the ORCHID project plans to develop the approach in ways that will enable more lives to be saved in the disasters of the future.