This paper focuses on the effect of categorization on historical tacit knowledge, in line with the TEI 2026 themes of “Data Interoperability" (or “Cross-Domain Application”). Within the constellation of my research, I propose to shift the center of gravity away from “Let’s categorize!” to “What happens when we shape memory this way?”. This study examines how applying core ontologies to the contributions of Canadian women of impact to history and archaeology, in an equity-aware approach, reshapes how cultural knowledge is modeled and shared across disciplinary boundaries.

While linked data standards are increasingly used in digital heritage infrastructures, the humanities data remain underrepresented and difficult to model in interoperable ways[1]. Despite large-scale initiatives and DH infrastructures, there is limited evidence on how existing models accommodate humanities-specific data types, especially those carrying tacit knowledge with them, such as personal archives and local place-based knowledge, and how these are encoded with minimal structural biases[2] [3].

This gap is addressed by using data from a set of archaeological archive collections held in Canadian institutions, as a testbed for assessing the possibilities and limits of modeling interoperable cultural data on material documenting heritage management activities. These documents include notebooks, correspondence, diaries, and extensive photographic materials capturing landscapes, artefacts, and field teams. Together, they record sites and objects but also research practices, collaborations, and relationships with the land and the communities over several decades.

How can models represent people, places, events, and objects in these documents in ways that respect their historical and cultural specificity while enabling cross-structures interoperability? How do we investigate the limits encountered when encoding and assigning persistent identifiers to entities drawn from multilingual, historically layered archival descriptions; which CRM extensions can accommodate concepts specific to specific times, languages, and cultures within a broadly shared ontology?

This project combines methods from qualitative data modeling with practical TEI encoding and RDF conversion workflows. It unfolds within a theoretical space structured by (1) critical semantic modeling, turning archival description into structured markup (what counts as “entity”, how to register uncertainty, contestation?), and (2) interpretive data modeling (how entities or their relationships will be modeled, understood, read). The analysis focuses on two intertwined dimensions: (1) semantic adequacy, where CIDOC CRM for example, provides a strong fit for events or artefact production and cases where it struggles (roles exceeding standard actors categories or nuanced distinctions); and (2) interpretive/ethical implications, as we adopt an intersectional perspective to examine how modeling choices inherited from archival descriptions, and authority structures shape who and what becomes visible [4] [5].

Preliminary findings suggest that ontologies (CIDOC CRM) significantly improve interoperability across documents, but the modeling process reveals persistent gaps: historically significant distinctions important to archaeologists and communities are difficult to represent with TEI, express in CIDOC CRM and map in RDF. Additionally, authority-based identifiers often reproduce existing asymmetries in whose names and places are standardized. These observations resonate with concerns in FAIR humanities data about the risk of losing thick description and about the dependency of researchers on external infrastructures and policies.