Digital Records, Algorithmic Tools - Reclaiming Historical Scholarship in the Age of AI -

This paper situates the shifts in the research environment of historical scholarship in the digital and AI era within the long-term historical process of transition toward “recording–search–algorithmic tools,” and explores historical research methodologies and their future directions. In human history, the emergence of new tools has always entailed both apprehension and opportunity; current AI technology similarly faces criticism for threatening human intellectual autonomy while possessing the potential to expand the horizons of research. The research environment of historical “big data” calls for a complementary integration of quantitative “distant reading” for the analysis of massive historical sources, based on the traditional qualitative “close reading.” Furthermore, the paper discusses the importance of data standardization and hub construction required for AI to process structured and unstructured historical big data in an integrated manner. In addition, it presents the HAVNet (Historical Archives Visualization Networks) database as a model of an active data hub for AI-based historical research. As empirical case studies, it introduces research on data reconstruction—linking individuals and households in the Daegu-bu Household Registers of Joseon dynasty using the RoBERTa model of Masked Language Models (MLM)—and on inferring the political affiliations of historical figures not explicitly recorded in sources through semi-supervised learning techniques. In conclusion, while AI technology maximizes the efficiency of historical analysis, the ultimate responsibility for source criticism and contextual interpretation still lies with human researchers. Consequently, researchers must establish a “researcher–algorithm loop” where algorithmic analysis and human interpretation form a virtuous cycle, thereby redefining the rationale for historical scholarship and education in the age of AI.