Semantic Analysis to Support Peace Analytics

Author: Roy Gardner

Peace processes are often defined and understood through the documents they produce, starting with peace agreements, but also including constitutions, implementing legislation, court decisions, and reports on implementation. Understanding peace processes as they unfold over multiple stages and arenas requires not just grappling with complex contexts, but also grappling with multiple forms of text, which is further complicated when attempting analysis over multiple countries and over long time periods.

This report describes the application of sentence-level semantic similarity technology for analysis of peace process documentation. Three projects were used to develop this work, which are traced throughout the report.

Peace Analytics Series

PeaceRep’s Peace Analytics Series features the research methodology underlying the PeaceTech innovations of the PeaceRep programme.

The series includes: data scoping research; ‘how to’ discussions relating to particular challenges in the field of visualisations and geocoding; and other proof-of-concept tech-based innovations, such as the use of natural language processing. It is intended to present the methodologies and decisions behind our PeaceTech digital research, to make it transparent, and to contribute to establishing a new research digital infrastructure in the field of peace and conflict studies, by supporting others to reuse and repurpose our methodologies and findings.