Introduction

Colonial-era moving image archives present a persistent metadata challenge: geographical names assigned by colonial administrations remain embedded in archival records long after independence, making collections invisible to researchers using contemporary geographical terms. The National Archives and Records Administration (NARA) holds over 520,400 moving image film reels, including substantial Korea-related content from the early twentieth century (NARA, 2024). A preliminary review reveals that many of these records carry Japanese-era place names rather than contemporary Korean ones. For example, the Korean capital appears as "Keijō" (京城 / 경성) rather than its post-independence name "Seoul" (서울). A researcher searching for "Seoul" will miss these materials entirely, while retrieving records tagged "Keijō" requires familiarity with colonial administrative vocabulary that most contemporary users do not possess. This study proposes an ontology network model to address this gap, with implications for other formerly coloniz ed countries facing analogous metadata problems.

Background

Metadata provides the structured description that enables discovery and contextualiz ation of archival resources (University of North Carolina Library, 2024). For moving images as cultural heritage, the Library Reference Model (LRMoo)—an extension of CIDOC-CRM aligned with the IFLA Library Reference Model—serves as the standard formal ontology for integrating bibliographic and museum information (CIDOC CRM SIG, 2024). While both CIDOC-CRM and LRMoo offer robust bibliographic description, neither accommodates the historical and political contingency embedded in colonial-era terminology. Postcolonial archival scholarship has documented this gap extensively (Stoler, 2002; Zinn, 1977 ), yet practical ontology-level solutions remain limited. This study addresses that absence.

Proposed Ontology Network Model

The study proposes an ontology network model that connects cognate geographical names—co-referential names drawn from pre-colonial, colonial, and post-colonial naming traditions—to their historical and linguistic contexts. Seoul illustrates the structure: the same territory was known as Hanseong (한성) before 1910, Keijō during Japanese colonial rule (1910–1945), and Seoul after independence. By linking these cognates through explicit temporal and linguistic properties, the model enables a search for "Seoul" to surface all historically co-referential records.

For moving image metadata, the model adopts seven CIDOC-CRM/LRMoo classes (F2 Expression, E55 Type, E19 Physical Object, E53 Place, F11 Corporate Body, F28 Expression Creation, E42 Identifier) and eight properties including P7 (took place at), P4 (has time span), and P48 (has preferred identifier). For cognate relationships, three additional classes (E4 Period, F12 Nomen, E56 Language) and four properties (R56 has related form, R35 is specified by, P4 has time span, P72 has language) are applied (CIDOC CRM SIG, 2024). A paradata layer—data about how metadata was produced and which names were selected—makes the political history of archival description explicit and traceable.

Methodology

The study proceeds in two phases. First, the ontology model is constructed using a NARA sample dataset and grounded in postcolonial and archival studies literature (Stoler, 2002; Baca, 2016). Second, the model's practical relevance is evaluated through semi-structured interviews with documentary producers and archival researchers who have worked with colonial-era moving image collections for over thirty years. Interview findings inform model refinement in two concrete ways: participants' accounts of search failures identify gaps the ontology must address, and their assessment of the cognate and paradata layers provides validation evidence for the model's fitness in professional archival practice. The model is designed to extend beyond Korea to other colonial contexts and languages, including collections from Africa, Southeast Asia, and the Middle East.

Expected Contributions

This study contributes a replicable ontology framework for improving semantic search precision in colonial-era moving image archives. By linking colonial and post-colonial geographical names through structured cognate relationships and surfacing the politics of archival description through paradata, the model increases accessibility for users unfamiliar with colonial terminology. The work advances postcolonial digital heritage practice and offers a transferable model for archival collections worldwide.