This poster presents an exploratory tool developed to analyze and visualize the nineteenthcentury correspondence of the influential Bulgarian merchant brothers Evlogi and Hristo Georgievi, whose economic, cultural, and philanthropic activities shaped Southeast European society and contributed to the founding of Sofia University (Mitev 2023). The project offers an integrated environment for exploring more than 680 TEI-encoded archival entries, enabling researchers to work with a dispersed historical corpus through spatial visualization, statistical exploration, and network analysis (Ahnert / Ahnert, 2019). In doing so, it shows how digital humanities methods foster connections between communities, places, and archival memory, aligning with the conference theme “Engagement.”

The Georgievi letters form a multilingual, transregional communication network linking Ottoman Bulgarian towns with the Danubian Principalities, Central Europe, and the Mediterranean. Greek, Bulgarian, Romanian, and French circulated together, reflecting the linguistic diversity of nineteenth-century commerce (Baramova 2023). Rather than relying on AI translation, the project uses human-encoded TEI metadata (TEI Consortium 2023) and philological attention to names, places, and commodities, showing translation, between languages, scripts, or archival formats, as a collaborative act. The tool lets users move across these multilingual traces, engaging with historical communication practices shaped by adaptation and negotiation.

At the project’s core is a commitment to annotation in a small data environment. Unlike largescale text-mining projects, our dataset is modest, because of limited source material, but it offers interpretative depth rather than broad trends. Each letter was encoded in TEI XML to capture correspondents, dates, geographical references, and standardized place names linked to a gazetteer. This highlights data as a historical artifact, not a neutral substrate. The resulting small but rich dataset shows how curated humanities data, especially in under-resourced languages, supports nuanced analysis while avoiding algorithmic opacity. By prioritizing human-driven annotation over automated extraction, the project encourages reflection on data creation, interpretation, and archival bias.

The Streamlit-based interface reflects these principles through three linked lenses. The map module uses Leaflet and Folium to visualize the network’s geographical reach, revealing its Lower Danube center and connections to Istanbul, Odessa, Trieste, Manchester, and London.

Filters trace how routes shift by correspondent, commodity, or time. The statistics view highlights high-level patterns, most notably Hristo Georgiev as the correspondence hub, while grounding each visualization in the underlying TEI records. Keyword facets connect micropatterns to individual letters, supporting movement between overview and close reading.

The network analysis module is the most innovative element. Built with NetworkX (Hadberg 2008), it constructs a directed, weighted graph modeling communication across the community. Node size reflects degree; edge thickness reflects correspondence frequency. The resulting hub-and-spoke structure centers on Hristo Georgiev, with clusters of strong ties and peripheral modules representing parallel ventures. This interactive exploration allows researchers to test hypotheses about commercial coordination, kinship, risk, or political organizing in ways that complement traditional historiography.

The environment also aligns with the sub-theme of “Remembering”. The Georgievi correspondence forms part of a cultural memory spanning multiple states and languages. Encoding it in TEI XML and making it available through transparent, FAIR workflows (Wilkinson et al. 2016) transforms archival documents into reusable, interpretable data. The tool amplifies rather than replaces archival memory, ensuring these letters, once locked in copybooks,

can be explored by scholars, students, and the general public. It also foregrounds provenance, authority control, and representational ethics, echoing debates in digital heritage.

By engaging with historical multilingualism, small-data annotation, and cultural memory, the project shows how Digital humanities approaches create new pathways between archives and contemporary audiences. Engagement here is relational: connecting past and present, distant correspondents and modern researchers, and individual letters with regional or transnational patterns. The newest version of the software, integrating expanded network analysis, has already helped prove hypotheses about trade goods, spatial relationships, and communication dynamics in the nineteenth-century Balkans. The tool is intentionally lightweight, reproducible, and extensible, enabling scholars, students, and cultural institutions to engage directly with the archive without specialized technical knowledge. Its design supports cross-collection comparison, inviting collaborations with other Balkan archives, minority communities, and regional heritage institutions. Future enhancements, such as temporal animation, topic modelling, and enriched semantic annotation, will help deepen the interplay between close reading and computational overview.

The project helps to illustrate how critical, interpretive, and responsible computational methods can deepen engagement with historical archives and prompt fresh perspectives on the networks, languages, and memories that shape the humanities.