The Edut 710 initiative, mobilized immediately following the October 7, 2023 attacks in Israel, represents an "archive of the present"—a distinct archival category built during an unfolding crisis. While rapid grassroots documentation captures raw, historically significant narratives, it inherently risks digital fragmentation, semantic ambiguity, and the loss of strict provenance. To address these epistemological vulnerabilities, this paper argues that Digital Humanities (DH) methodologies must serve not merely as post-hoc analytical tools, but as the constitutive infrastructure of the archive. By implementing a relational data model and linked entity framework based on an ontology of Testimony, Person, and Location, the project structurally enforces consistency and secures the chain of custody while embedding ethical principles directly into the architecture. Furthermore, integrating a Retrieval-Augmented Generation (RAG) framework enables "distant listening" at scale, uncovering macro-patterns across over 1,700 hours of testimony. Ultimately, treating data modeling as an ethical practice ensures that computational insights remain anchored to personal, localized experiences, future-proofing the living archive while serving the documented community.

Historical archives, once digitized, generate vast datasets that remain largely unused, representing a significant opportunity for interdisciplinary research in the field of Digital Humanities. This work focuses on the organization and structuring of the medical records of the former psychiatric hospital of Reggio Calabria (Italy). Central to this process is the use of metadata, which provides a framework for describing, contextualizing, and linking digital documents to ensure meaningful access and interoperability. To maximize the value and usability of the medical records of the former psychiatric hospital of Reggio Calabria and enable their retrieval and reuse for research purposes, it is necessary to combine structured data models, standardization practices, and advanced text recognition technologies. This integrated approach transforms the digitized medical records into a resource suitable for historical, medical, and social science research. The study’s main scope is to explore ways to structure these documents while preserving their historical and structural/formal characteristics.

Digital humanities research software and infrastructures are typically designed around highly specific scholarly requirements. While effective for particular research practices, they often lack transferability across domains. The Text+ Registry, developed within the Text+ consortium of the German National Research Data Infrastructure (NFDI), proposes an alternative approach. Rather than serving solely as a catalogue for editions, lexical resources, and collections, it is conceived as a reusable infrastructural component adaptable to diverse scholarly and institutional contexts.

This paper examines three architectural principles underlying the Registry: metamodelling, model-driven design, and information layering.

Furthermore, two application scenarios demonstrate broader reuse: a transnational meta-archive of German archival materials on modern India (MIDA) and a planned NFDI-wide helpdesk catalogue for research data management services.

Technically, the Registry harvests distributed data via standard protocols while aligning heterogeneous schemas through configurable mappings.

The Text+ Registry thus models an infrastructural philosophy that supports pluralism: different conceptual frameworks can coexist and interoperate without enforced standardization. This approach reconciles disciplinary specificity, interoperability, automation, and curation, offering a sustainable model for digital humanities infrastructures and potentially for research infrastructures beyond the humanities.