This poster presents a project reconstructing the information network of scientific publishing in Czechoslovakia (1910–1938). It maps the development of science in a newly established state facing tensions between national independence and international trends. Integrating Digital Humanities and Information Science, the project views science as a complex process of knowledge circulation, overcoming traditional biographical narratives through macroscopic modeling.

Methodologically, the research relies on Social Network Analysis (SNA) and distant reading of a large corpus of digitized scholarly periodicals. The application of infometrics and specifically SNA allows for an examination of scholarly communication as a macroscopic model of thematic, institutional, and personal relationships. The research addresses three methodological challenges in processing historical texts:

- Automated document segmentation of OCR data lacking structural metadata.

- Named Entity Recognition (NER) for identifying authors and institutions within graph databases (Neo4j).

- Comparing traditional (LDA) and neural (BERTopic, LLMs) topic modeling for interpreting historical scientific terminology.

The poster presents a pilot data sample through an interactive demonstration. Visitors can explore the modeled network, visualize topic transformations, and discuss the limitations of automated OCR processing in historical research. This study provides critical insights into the interplay of culture, politics, and scientific discourse.