2025 - November and December - page 30
Image details
| Issue number | 404 |
|---|---|
| ISSN | 2632-7171 |
| Publication date | 1st November 2025 |
| Transcription |
magazine Industry News Visual Discovery and Structured Data: AI-Assisted Access to Collections Over the past three years, innovative archives have tackled a common challenge: turning large, under- described holdings into collections people can actually find and use. Two patterns stand out. First, data extraction—converting scans into structured records so staff can search, link and reuse. At the Swiss Federal Archives, tens of thousands of Prisoner of War (POW) index cards became searchable records within weeks rather than months or years, because archivists set the quality bar and let AI handle the repetitive work. After review, results are exported into existing systems— preserving workflows and standards. Second, visual discovery—making image collections searchable without manual item-level cataloguing. At Galt Museum & Archives (Canada), 18,000 historic newspaper photos became instantly browsable through a “find-by-image” search [link https:// galt.archipanion.com], enabling staff and the public to jump from “looks like this” to relevant matches even when metadata is sparse. In Bern, the Swiss Post, Telegraphy and Telephony (PTT) Archive and Museum of Communication made 40,000 digitised images easily discoverable and the Swiss Federal Archives also made 200+ hours of film accessible, with search results that open at the exact WWII PoW Card Swiss Federal Archives © Swiss Federal Archives, Archipanion moment in the footage. A practical next step: pick a safe subset, define success and quality checks, run a short pilot, and measure. The pattern is consistent—faster access, lower backlog, high standards. Markus Stauffiger, Director, Archipanion Visual Search - Galt Museum & Archives © Galt Museum & Archives, Archipanion 30 |