Extracting Named Actors from Text: Using Named Entity Recognition (NER) in Peace and Conflict Studies

Abstract

This report explores the potential role of Named Entity Recognition (NER) in peace and conflict studies, drawing insights from PeaceRep’s use of the technology to extract signatory organisations from a large-scale database containing peace agreements – the PA-X Peace Agreement Database. PeaceRep’s successful application of NER in extracting signatory data from peace agreement texts provides an example of how to navigate complex textual data relevant to ending conflict, in a time where we increasingly require data as evidence for decision making. As text remains the primary medium in this domain, NER emerges as a crucial tool in identifying, extracting, and structuring key data, facilitating streamlined research and information extraction.

Highlighting NER fundamentals and methodological approaches, this report advocates for transparent artificial intelligence, human oversight, and a functional approach tailored to peace and conflict studies. Accompanied by practical demonstrations in Python Jupyter Notebooks, the report showcases NER’s applications—from document summarisation to geospatial and temporal analysis. Ultimately, it emphasises NER’s potential in deciphering complex textual data while emphasising, and recommending, the need for nuanced approaches that balance NER’s strengths with its limitations in this critical field.