The paper proposes a digital approach to analysing cultural data in the context of educational activities. Automation of data analysis allows supplementing research performed manually by scientists, especially in conditions of large samples. The area of its application, that is, cultural data analysis for educational purposes, contributes to preserving historical memory and simplifies the learning process in terms of knowledge presentation. The authors select digital copies of Armenian manuscripts as the data under study. The problem of direct recognition can be defined as cultural stratification based on ancient manuscripts in such areas as identification of various kinds of material and intangible culture, typology of people’s cultural activities (for instance, identification of various types of professional, religious, linguistic, regional culture or culture associated with historical stages of society development). The work considers text recognition and accuracy assessment, formulates hypotheses for analysis that allow studying aspects of historical sources. Based on these hypotheses, the authors make conclusions on linguistic and cultural segmentation and requirements for the ergonomic display of the cultural data analysis.