Researchers have developed a computational text-analytic framework to quantify cultural-psychological constructs like individualism, collectivism, and honor within historical Ottoman Turkish court records. By adapting sentence-embedding models to the specific language of these 54,670 documents, spanning from 1513 to 1911, they created a pipeline to analyze cultural variations. The models demonstrated strong performance in document retrieval and semantic similarity tasks, with an adapted sentence transformer achieving a Recall@1 of .843 and a Mean Reciprocal Rank (MRR) of .806. The study's findings were further validated through expert annotations and qualitative analysis, revealing historically specific legal and institutional patterns. The developed models, representations, and scoring pipeline are being made publicly available to facilitate reproducible historical-psychological analyses. AI
IMPACT Enables new methods for studying historical societies and cultural change through advanced text analysis.
RANK_REASON Academic paper detailing a new computational framework for historical text analysis. [lever_c_demoted from research: ic=1 ai=1.0]
AI-generated summary · Google Gemini · from 1 sources. How we write summaries →