The future of human rights investigations: Using open source intelligence to transform the documentation and discovery of human rights violations

Lead Research Organisation: Swansea University
Department Name: School of Law

Abstract

Technology is rapidly transforming how investigations of human rights abuses are carried out. Traditionally, investigations relied upon witness testimony and on-site evidence to prove the existence of human rights violations. More recently, however, human rights investigations have been turning to Open Source Intelligence (OSINT), such as social media content and satellite imagery, to overcome the physical, security, and societal barriers to gathering reliable evidence. In August 2017, the International Criminal Court issued its first arrest warrant based on social media evidence.

OSINT has the potential to democratise the flow of information on international human rights violations in an unprecedented way. By allowing investigations to be carried out remotely, and by enabling information to be received directly from witnesses and victims rather than through intermediaries, OSINT can break down some of the barriers that have silenced some voices in traditional investigations and prioritised others. However, new issues arise with these types of investigations. The huge volume of evidence retrievable from social media can make it difficult for investigators to extract truly useful information. There are further issues of informational bias that can be attributed to algorithmic bias or to misinformation posted online, intended to obfuscate or exaggerate human rights abuses.

By combining a unique multidisciplinary methodology, drawing on socio-legal, computer science, and geospatial analysis methods, this project asks:
"To what extent can OSINT be leveraged to contribute more systematically to human rights investigation and documentation? Can natural language processing and geospatial methods for analysing social media content assist in the discovery and analysis processes, and help overcome potential issues of informational bias and misinformation that may arise?"

It will:
1. Create the first ever overview of the use of OSINT by UN human rights fact-finding missions. Through interviews with members of UN Commissions of Inquiry and human rights investigations (many of whom we have worked with on other projects) and a project workshop, we will identify the barriers and reservations to their use of OSINT. Combining this data with a systematic review of reports produced by these investigations, we will determine the extent to which information gathered through OSINT methods could address some of the informational gaps inherent to traditional investigative methods.
2. Develop, in collaboration with human rights organisations, the Knowledge Hub Framework (KHF), a set of core microservices that will provide tools to gather data and carry out specific analytical tasks, such as comparing documents for similarity, identifying place names within free text and mapping them, and assigning weightings and confidence ratings to data sources based on automated crosschecks, validations, and historical accuracies.
3. Through the KHF, use natural language processing, text mining, and spatial analysis techniques, combined with legal analysis, in a case study to demonstrate how OSINT-based investigations could be made more systematic. Our case study will focus on The Philippines, where mass human rights violations have allegedly occurred, but which is not currently subject to a UN human rights inquiry, and which has witnessed a proliferation of social media accounts spreading counter-narratives about alleged human rights abuses. In a dedicated workshop, we will demonstrate the prototype KHF to stakeholders. We will later offer training sessions for human rights organisations. The Institute for International Criminal Investigations has agreed to host one such training session in The Hague.

As well as the KHF, which will be updated as new functionalities are created, the project will result in three academic journal articles and a Guide to OSINT for Human Rights Organisations. It has the potential to transform human rights fact-finding.

Planned Impact

A. United Nations Commissions of Inquiry, Human Rights Fact-Finding Missions, and Human Rights Non-Governmental Organisations
These organisations are increasingly denied access to affected regions (Kirby, 2014) and as such, are more likely to depend on Open Source Intelligence (OSINT) in the future. They will benefit from our research on this project in three important respects:
1. By mapping the space in which OSINT is used by UN human rights investigative missions, our research will make these organisations aware of the potentials of OSINT;
2. By highlighting the potential issues of informational bias in the use of OSINT, and determining whether systematic methods, as pioneered in our case study, could partly overcome those issues, our research will increase awareness of these issues;
3. Our Knowledge Hub Framework tool will be a useful for these organisations, by automating some of the discovery and verification processes that are currently carried out manually. The Knowledge Hub Framework will be developed in collaboration with organisations that we work with (namely, the Institute for International Criminal Investigations, the Office of the High Commissioner for Human Rights, and Amnesty International). It will be made available for use to them and to other human rights organisations, to whom we will give an access key.

We will ensure that these organisations have the opportunity to benefit from our research through:
1. Publishing three open access journal articles on the use of OSINT by human rights fact-finding missions, which will be widely distributed;
2. Presenting our findings and offering training and support on the Knowledge Hub Framework to OHCHR staff in Geneva and members of the Syria Monitoring Team in Beirut;
3. Interviewing members of missions, and inviting them and other OHCHR representatives to participate in project workshops;
4. Widely sharing the "Guide to OSINT for Human Rights Organisations" that we will prepare, and engaging with news media on our findings.

B. International Criminal Tribunals
These courts are beginning to use OSINT in their prosecution of international crimes (ICC, 2017). There is an acute need to identify and start to overcome issues of informational bias, given that the investigations are increasingly likely to depend on OSINT. In addition, our Knowledge Hub Framework will be a useful tool for investigators. We will conduct training on the tool with the Institute of International Criminal Investigations (IICI) in The Hague, which has agreed to host this training. As with (A) above, we will ensure that these tribunals are provided with a copy of our open-access journal articles and Guide to OSINT, and that staff are invited to participate in workshops. CI Koenig is Chair of the International Criminal Court's Office of the Prosecutor's Technology Advisory Board, providing an invaluable link to the Court, and she will transmit our findings to the Court.

C. Victims of Human Rights Violations
Ultimately, we hope that our findings will lead to enhanced accountability for human rights violations through stronger investigations. We are aware that this may seem like an ambitious goal, but it has been seen in recent situations like Myanmar that, through OSINT, human rights violations that would otherwise have been hidden can be revealed. Our findings will highlight how more systematic methods of gathering and verifying OSINT can be utilized, including through the use of our Knowledge Hub Framework tool, to which human rights organisations will be given access. Our case study on The Philippines will not only provide a test case for the development of the Framework tool and a means to demonstrate its functions, but will also potentially result in information on whether human rights abuses have been committed in The Philippines. We will share our with the International Criminal Court through a briefing paper, if it seems that international crimes may have occurred.

Publications

10 25 50