THE EMPEROR, HITCHCOCK AND PATRIARCHY: AI-ASSISTED VS. HUMAN GENERATED KEY WORDS EXTRACTED FROM UNPROCESSED ETHNOLOGICAL MATERIAL

Апстракт

Archives of audio, video, and written ethnographic materials produced through individual field research frequently suffer from inadequate or inconsistent keyword classification, which can obscure their thematic content and hinder effective retrieval. This article explores the extent to which computational information systems can support professionals in improving the organization and accessibility of such archival materials.

The study applies a selection of computationally intensive methods for text search and clustering to a corpus drawn from the Digital Archive of Ethnological and Anthropological Resources (DAEAR) at the Institute of Ethnology and Anthropology, Ss. Cyril and Methodius University in Skopje. The practical usefulness of these methods is assessed through a comparative analysis of machine-generated classifications and those produced by human evaluators. Specifically, the results of automated extraction and clustering are compared with classifications provided by ten experts (ethnologists and anthropologists) and ten non-experts.

By examining convergences and divergences between algorithmic outputs and human judgment, the paper reflects on the epistemological and practical implications of integrating computational tools into ethnographic archival practice. The findings contribute to ongoing discussions on digital humanities methodologies and the management of qualitative research data in the social sciences.