Five teams of researchers presented data-driven solutions to societal and environmental problems, having just completed a 12-week Data Science for Social Good Fellowship Programme, at the end of August.
The projects centred on; intervention for predicted high-utilisers of emergency services in Memphis, USA, reducing the time it takes to answer legal questions in Uganda, measuring traffic to reduce air pollution in the UK, improving cardiograms and improving prospects for jobseekers in Portugal.
Each team of three or four fellows studying at universities across the globe works with a technical mentor, a project manager and a partner co-ordinator. One project aimed at improving legal access in Uganda brought together data scientists from the University of the Andes, the University of Illinois at Urbana-Champaign, Virginia Commonwealth University, Sciences Po and University College London.
The problem they were addressing was the lack of access Ugandans with legal problems have to lawyers and legal advisers. The vast majority of lawyers live in the capital Kampala, while the vast majority of Ugandans live outside of the capital. Working with Barefoot Law, a non-for-profit organisation that provides free legal advice through mobile technology, the fellows sought to reduce the amount of time it takes to provide a beneficiary with a legal answer which was previously 72 hours.
The fellows decided to use artificial intelligence by feeding historical data into machine learning algorithms to automatically draft responses which would be vetted by qualified lawyers, aiming to reduce this time to less than 24 hours.
They framed this as an information retrieval problem, where questions were treated as queries and historic Q&A pairs were documents to be retrieved. Through a process of pre-processing, feature generation, prediction and deployment, the team produced a model that could generate more accurate answers for simple questions. This resulted in reduced response times, lawyers having their time freed up to address more complex cases and an improved data strategy.
The programme was run by the Gandhi Centre for Inclusive Innovation at the Imperial College Business School, in collaboration with the University of Chicago. This is the seventh year that Data Science for Social Good programme has been running and the first time a programme was held in the UK. There were two in total; one at Imperial College London and another hosted by the Alan Turing Institute in conjunction with the University of Warwick.
Professor Nick Jennings, Vice-Provost (Research and Enterprise) said: “Getting excellent research into the public domain and making a difference to society is something Imperial is passionate about, and our mission aligns well with this fellowship scheme.”
Dr Sankalp Chaturvedi, director of the Gandhi Centre, said: “The Data Fest event was the perfect way to showcase how the Gandhi Centre is undertaking more enterprises with a social and inclusive impact. This was a wonderful opportunity to host researchers from all over the world.”
Data Science for Social Good is based on three themes: education, incubation and collaboration. In addition DSSG focuses on three values of being open, collaborative and ethical. The goals of DSSG are to train fellows, train governments and non-profit organisations and to build communities.