INTEGRATING ARTIFICIAL INTELLIGENCE INTO COMPARATIVE BIBLIOMETRIC STUDIES: FAMILY REPRESENTATIONS IN DOCTORAL RESEARCH (TURKEY–GERMANY, 2019–2024)

Abstract

This research expands upon a previous comparative analysis of family representations in doctoral dissertations from Turkey and Germany, (Cöre Güzeller & Pöhl, 2020) through the inclusion of artificial intelligence (AI) in the bibliometric and thematic analysis processes. More than 300 dissertations (2019–2024) from the Turkish Council of Higher Education Thesis Center and from the Deutsche Nationalbibliothek form the corpus of research. A mixed-methods research framework, assisted by AI, is used which utilises large language models (LLMs), cross-lingual semantic analysis and human-controlled validation. Using natural language processing and topic modelling, the AI identifies underlying thematic clustering such as value transmission, family-school collaboration and migration and integration, as well as mapping their disciplinary distribution. Further, a multilingual embeddings model (Sentence-BERT) allows for analysis of semantic proximity between discourses on the family in Turkish and German language research, revealing common and differing conceptual trends. The study finds that AI-supported semantic clustering supports the depth, rapidity and intercoder reliability of comparative bibliometric processes, found by the high Cohen’s κ scores obtained against manual coding. The study finds that Turkish dissertations see the family as primarily a pedagogical and sociocultural phenomenon, while German dissertations emphasise psychological and social work approaches, particularly in terms of migration and inclusion context. The use of artificial intelligence in partnership with human interpretation results in the creation of an AI-assisted model for comparative educational research, supporting evidence-based policy making and the Türkiye 2025 "Family Year" initiative.