Background and Objectives

The archival description standard ISAD(G) has popularized hierarchical description using fonds/series/file/item models, and systems such as ArchivesSpace and AtoM are designed based on this basis (ICA 2000; Bushey 2012). By contrast, many university archives in Japan employ local call numbers that combine department, year, and volume into a single string, thereby uniting physical and logical units. This practice complicates straightforward mapping to the standard hierarchy and is seldom discussed on an international scale.

Tohoku University Archives manage a diverse collection of institutional records, personal and affiliated-organization records, and university publications using these local call numbers. These materials are made accessible through the Tohoku University Digital Archives (ToUDA) (Hanzawa 2024; Kato 2025) and collectively serve as crucial evidence for the university’s institutional, intellectual, and social history. However, the current system faces challenges as the call-number-based hierarchy is difficult to represent visually online and users accessing ToUDA are required to submit separate paper application forms for consultation or reproductions. The discovery → delivery chain (Dempsey 2006) is thus fragmented; the combination of varied call numbers and paper-based processes hampers access for university historians, alumni, and the general public.

To address these issues, we are re-evaluating the data model, search interface, and access request workflow for the entire collection. This poster presents an evolving design that aligns a call-number-based hierarchy with an integrated discovery → delivery workflow and suggests reusable patterns for contexts poorly served by ISAD(G)-oriented systems, especially those using compact, string-based call numbers in East Asian university archives.

Methods

We started by analyzing sample call numbers. For instance, in “General Planning / 2006 / 31-1 / Item 006,” “General Planning” indicates the transferring unit, “2006” indicates the transfer year, “31-1” represents the file, and “Item 006” refers to an item within that file; future units, such as “Item 006 / 1,” are anticipated. The first two segments serve as facets of transferor and year, while “31-1” and “Item 006” establish the effective file and item levels.

Based on this, we decomposed call numbers into components such as department, year, file number, and item number. We extracted the transferor and year as separate facet attributes, while segments, such as “31-1” and “Item 006,” function as hierarchical nodes featuring parent IDs and hierarchy levels. In close collaboration with archival staff and external system developers, our goal is to maximize the reuse of existing call numbers while avoiding disruptive changes or rekeying. The resulting schema models a variable-depth hierarchy that accommodates future depth extensions by combining parent-child relationships and category types, while assigning stable URIs to each unit (Fig. 1).

Figure 1. Decomposing a local call number and mapping segments to facet attributes and hierarchical nodes

Regarding the search interface, we maintain a flat list view that allows records to be sortable by fields such as ID or date, rather than grouping or indenting child records in the results as this would impede flexible sorting. Instead, the list itself acts as an entry point into the access workflow: users can select multiple records through checkboxes in the results or detailed views. The selected IDs and basic metadata are temporarily stored in the browser and automatically filled into a predefined application form template. This functionality enables users to create application documents with item information prepopulated, thus reducing user effort and minimizing follow-up corrections needed by the institution.

Current Insights and Future Work

Viewing local call numbers as a blend of facet-like elements (transferor, year) and hierarchical elements (file and item levels) offers a pragmatic approach to clarify the on-screen hierarchy without altering the existing classification system. As the separation between explicitly stored hierarchy and structures derived from call numbers directly influences input requirements and system complexity, we adopt a “minimal” strategy: relying on structures that can be automatically inferred from call numbers while storing limited redundant metadata such as parent IDs, hierarchy levels, and facet attributes such as transferor and year.

Future work will include pilot implementation and evaluation of the proposed design along with its integration into an RDF-based model of organizational change and AI-assisted search driven by natural language queries. Through these advancements, we aim to enhance access to Japanese-language records documenting university organization and contribute a design case rooted in a Japanese university archives context to discussions on archival practices within digital humanities, which have thus far primarily focused on Euro-American, English-language examples.