Digital Shezhere: AI-Assisted Reconstruction of Oral Genealogies in Central Asia

Abstract

Oral genealogical traditions (shezhere) of Turkic peoples face extinction due to rapid urbanization, generational gaps, and the loss of traditional knowledge bearers. Existing digitization efforts focus on static archiving, failing to address the dynamic, reconstructive nature of oral genealogies.

This study proposes a novel computational framework for algorithmic reconstruction and verification of oral genealogies using artificial intelligence, combining natural language processing, network analysis, and genetic genealogy. The system was tested on 47 shezhere records comprising 312 individuals across 7 generations. The algorithm achieved 84.6% accuracy in reconstructing missing genealogical links and 91.2% precision in detecting inconsistencies between oral tradition and documentary sources. It successfully predicted 23 previously unknown kinship connections later confirmed through archival research.

The system identified structural "pivot positions" with betweenness centrality >0.7, revealing the functional nature of the 7th ancestor position. AI-assisted reconstruction preserves the living character of oral genealogies while providing verifiable accuracy, enabling scalable digitization of endangered traditions across Central Asia.