Using Data Science to Examine Conflict-Related Sexual Violence

The lack of data is a major barrier to comprehending conflict-related sexual violence (CRSV). Existing studies indicate a high prevalence of violence but limited documentation. Media are often the first to report CRSV, thus, can data science methods analyze news reports promptly and accurately? We compare manual, machine-learning and Generative AI analysis of thousands of media reports. Five variables can be automatically coded with high accuracy, increasing to over 90% with "chain of thought" prompting. Data science can reveal previously unknown CRSV attributes to inform timely prevention, but it has important limitations that researchers, advocates, and policymakers should be aware of.