April Blog post

As an accompaniment to York’s #ArchiveApril posts, we thought we would take this opportunity to expand on the series of events and ideational processes that led to the creation of the xDX Project’s bespoke archival model. As mentioned in our February blog post, a huge swath of this work was conducted over the course of our 5-hour train ride back to Toronto in May 2024. This energetic whirlwind of ideation, prototyping and the resulting archival model were a direct response to the rich discussions that took place during the xDX Workshop II, co-hosted by the Carleton xDX team and our partners at the Canadian Museum of History. The Workshop came on the heels of several immensely productive activities and events, such as the three-day “Forward Linking” Conference hosted by LINCS, which brought together digital humanities experts from across Canada to discuss the more than twenty projects that are currently under development.
‍
The projects, and the research teams responsible for them, provided us with new insights into the possibilities of semantic web technologies. Rosalind’s subsequent participation in the 2024 Digital Humanities Summer Institute amplified these new learnings. The (then-)newly brokered collaboration between the YorkU xDX Project and the Clara Thomas Archives and Special Collections (CTASC) provided us with the ‘test case’ we needed to operationalize our new understanding. This agreement marks a significant accomplishment for the xDX project, ensuring ongoing stewardship of our 106 artefacts and the Fred & Glenn Moffatt Fonds by York University Libraries. The earliest version of the xDX model, pictured below, was borne out of several workshopping sessions in 2022 and 2023 lead by our infrastructure partners at LINCS and CHIN with the following goals in mind: To reconnect the now dispersed artefacts and archives that made up the Design Exchange (DX) To enhance discoverability of artefacts and archival materials across partner institutions To generate possibilities for the discoveries of unanticipated connections and novel uses by researchers.

‍

The xDX model employs the CIDOC-CRM ontology. CIDOC-CRM was developed specifically for the field of cultural heritage and operates as the framework that makes our open-access data machine-readable and, by proxy, both query-able and editable to researchers. The development of this model would not be possible without the assistance of our partners at LINCS (Susan Brown, Kim Martin, Sarah Roger, Natalie Hervieux, Jessica Ye and Dani Metilli) and CHIN (Philippe Michon and Trang Dang Nguyen Khanh).

While the original model was intended to reconnect the physical artefacts of the former DX collection via Linked Open Data, we soon realized that we needed to head back to the drawing board to devise a data model that would allow us to integrate the xDX archival materials held at York, Archives of Ontario (AO), the Canadian Museum of History (CMH), and the Royal Ontario Museum (ROM). After some deliberation, we decided that rather than creating individual data points for each and every document—which ranged from correspondence and newspaper clippings, to photographs, technical drawings, and layouts—we decided to employ archival cataloging practices, focusing on the file level, rather than each specific item. This decision not only allowed us to reduce the number of newly minted records, from thousands to hundreds, but also highlighted a critical tension between certainty and ambiguity in our dataset(s). Perhaps counterintuitively, we’ve found that allowing for uncertainty, rather than insisting on precision, has resulted in a data set that is fertile ground for innovative research questions and novel connections across partner collections.
‍
‍

Rosalind Sweeney-McCabe’s diagram of the xDX Archival Model.

Mapping archival materials at the file level allowed us to develop not only multiple connections but also multiple types of connections between entities (people, companies, places, objects, etc.), without needing to precisely define these relationships in advance. Using the affordances of CIDOC-CRM, we opted for two connecting terms (referred to as properties in LOD): “is about” and “refers to”. In our model, “is about” is used to indicate a file’s specific subject (when available) and “refers to” operates as a more broad, unspecified connecting point between various people, places, or groups and the file.
‍
‍

Rosalind Sweeney-McCabe’s experimental diagram of Steamer archival connections.
‍

The image above illustrates an experimental mapping of our model. This hypothetical Archives of Ontario file “is about” the Steamer Chair Series, which was designed by Thomas Lamb and manufactured by several different companies over the course of its production timeline. It ‘links’ various steamer chairs (actual artefacts) held by different partner institutions, alongside several related archival files. The designer, Thomas Lamb, is also directly linked to this file, allowing for discoverability of all Lamb-adjacent materials when queried. Under “refers to” more tangential connections are made searchable, such as connections to photographs of Steamer factories in Malaysia or other corollary materials, such as letters from MoMA, discussing the inclusion of Steamer Chairs in their collection. Flagging relationships, as opposed to directly defining connections, affords scholars the opportunity to explore and discover their own paths through the materials and (we hope!) encouraging them to visit the institution where the materials and artefacts are housed.

Ultimately, through this work, we have found that our collection of industrially-designed objects and archives intersects with principles of linked-open data in an intriguing fashion, simultaneously challenging the traditional limitations of collection management systems and centering the agency of future researchers. The very essence of our collection of mass-produced artefacts— which includes both pristine and well-used models, prototypes and molds, sketches and pressproofs—embodies the entanglement of values of experimentation, iteration, and ambiguity, effectively replicating the essence of our collection in the model itself.

Jan Hadlaw, xDX Project PI
Aviva Weizman, Lead RA
Rosalind Sweeney-McCabe, MA RA
York University