Documenting potential Human Rights Violations using Natural Language Processing, Images and Open-Source Information

The project has continued work on developing new approaches to documenting human rights through the use of natural language processing, images and open-source information. In January Dr Grigorios Kalliatakis, Dr Shoaib Ehsan, Prof Ales Leonardis, Prof Maria Fasli and Prof Klaus McDonald Maier published an article in IEEE Access on ‘Exploring Object-Centric and Scene-Centric CNN features and Their Complementarity for Human Rights Violation Recognition in Images.’ This article explores how potential abuses of human rights could be inferred from images. In this work, the notion of human rights violation recognition as an area of practice within computer vision was first announced, by developing the first ever purpose-built human rights violation classification schemes based on deep image representations. The Human Rights Archive Database was also introduced, a verified-by-experts repository of approximately 3K human rights violation photographs, labelled with human rights semantic categories, comprising a list of the types of human rights abuses encountered at present.

Prof Lorna McGregor also published an article in the AJIL in 2019 entitled ‘Are New Technologies an Aid to Reputation as a Disciplinarian?’ This article looks at the relationship between the reputation and accountability of international organisations (IOs) through the prism of new technologies. Lorna argues that these new technologies open routes to accountability that may lead to direct accountability as well as a structural analysis of how to prevent violations in the future. She then goes on to identify factors that may cause an IO to deem its reputation to be at risk and therefore respond to a given situation.

Important work in this field of documenting potential human rights violations is also being carried out by the Digital Verification Unit, which is affiliated with the HRBDT Project. The unit works to document and verify digital evidence pertaining to human rights abuses, and to use open source investigative techniques in the pursuit of accountability. HRBDT researcher, Dr Daragh Murray and consultant Sam Dubberley head up The Digital Verification Unit which forms part of Amnesty’s Digital Verification Corps. In November 2019 this partnership between Amnesty International, not-for-profit company Airwars and six universities from around the world, including Essex’s Digital Verification Unit, was named International Collaboration of the Year at the Times Higher Education Awards.