Digital Mapping to Protect Freedom of Religion or Belief
THE QUESTION
Can AI successfully monitor and capture trends in Freedom of Religion and Belief and inform the UK government’s actions on FoRB?
LOCATION: United Kingdom
SECTOR: Government
TECH: AI/ ML
TIMELINE: September 2022 - Present
PARTNERS: Blue Globe Innovation
IDEA STAGE: Discovery, Pilot
The Challenge
Violations and abuses of the right to freedom of religion of belief (FoRB) are hard to track and often hidden amongst other forms of human rights transgressions. Data is largely anecdotal and of variable quality and there is insufficient analysis of the data to provide a sufficiently clear picture of the shape and scale of the problem in real time, either to inform appropriate interventions or to analyze progress.
The Idea
While some work has been done on open source evidence collection, the project proposes to use AI/machine learning algorithms, trained to identify FoRB terminology and FoRB related risks using natural language processing, to analyse online open source FoRB related data, and plotted onto GIS (geographic information system) visualisation software, to generate a heat map of places where FoRB is under threat.
This real time visualisation of FoRB related data will integrate information from a range of sources and present a dynamic display about the changing situation, enabling policy makers to see both positive and negative developments on the enjoyment of FoRB mapped geospatially. If successful, this pilot will enable better use of data, to explore synergies across different data and enable those seeking to promote and protect FoRB to better target interventions.
Key metrics
Harassment of religious groups is reported in more than 90% of countries according to Pew Research 2018 data.
Stakeholders, including academics, civil society and other interested governments confirmed that there were no existing tools which brought together information on FoRB in this way.
Our research showed that there were other platforms that used AI and machine learning but none focused on gathering and analysing FoRB data.
What we learned
While Pew Research Center gathers data to analyse religious change, it does not use AI or publish in real time. There are no platforms tracking FoRB but there are several comparable platforms which use machine learning (with manual validation) to track other related issues, e.g. on conflict early warning, civic space, and analysis of hostile government narratives.
The proposed tool would gather data from a small number of online news sources. A more maximalist approach would gather data from a greater range of sources (e.g. traditional media, reports, court documents, legislation, local surveys, social media). This would involve using customised defined keyword filters to scrape data, which could then be refined using natural language processing and machine learning techniques. Some manual validation would be required. A ready-made solution for visualisation of the data is preferred to building a reporting tool from scratch.
Data sources that can be considered comparable platforms include news reports and global news outlets. We identified several data collection tools, including the Dow Jones Factiva aggregator, BBC Monitoring and Lexis Nexis.
What happened next
With the learnings produced in this discovery stage, the team would like to look into the current experience and challenges faced by the target user groups within FCDO. This would expand understanding around their needs from this platform and would allow the team to redefine the expectations from such a platform and how it would be used.
The team also realises that there are still several unanswered questions around data sources and how the data would be analysed before development of the AI model can be considered. They are looking to secure further support from the FT Hub to scale up to a pilot which will provide additional funding and support to look into the above areas and to develop an MVP.