Meet the winners of the 2026 APE Digitisation Grant
Italy - Franco Cordero: Reorganization and AI-based description of the archive
Estonia - The Vaeküla Manor
The Vaeküla Manor (Vaeküla mõis) is a historic Baltic German landed estate in northeastern Estonia, in Lääne-Viru County, near Rakvere. It was first documented in 1532, and it originally belonged to the network of aristocratic agricultural estates that characterised Estonia under Baltic German landownership. Today, its manager Niina Nihtinen was awarded the APE Digitisation Grant to digitise its archives, and to restore Vaekula "as an internationally-connected centre of cultural memory, research, education, and dialogue accessible to future generations", in the words of Niina. Niina is an entrepreneur and owner of the Vaeküla Manor. She says: "I regard ownership as responsibility to protect a hidden European Heritage. I have driven extensive research, engaged students, and built international partnerships across Estonia, Sweden, Germany, and Switzerland to uncover records dating back to 1490. Our findings highlight Vaeküla`s significant historical connections to Sofia Kovalevskaya and Schubert family of Vaekula"
Italy - Franco Cordero: Reorganization and AI-based description of the archive
Versione italiana disponibile qui
The digital archive of Franco Cordero, an important jurist who gave fundamental contribution to criminal procedural law,consists of born-digital documents stored on the computer used by the scholar: primarily text files and, to a lesser extent, images and emails. The organisation of the digital documents largely reflects the folder structure found on the computer.
Turning his personal computer into an archive was a challenge for the Polo del Novecento , but an important one as more and more historical archival collections will be in this format, going forward in archiving the 21st century. With the APE Digitisation Grant, the Polo will conduct:
- Archival survey and description, including AI-based analytical cataloguing.
- Arrangement and structuring of the archive, using an AI system to automate cataloguing and content searching.
- Import of descriptive catalogue records into 9centRo Vivo, and export of descriptions in EAD/CSV format for ingestion into the Archives Portal Europe (APE).
The resulting platform is available here: 9centRo Vivo – Polo del '900.
Before proceeding with the arrangement and inventorying of the archive, it may be necessary to carry out a selection process, removing empty folders, duplicate files, folders containing software operating instructions, and potentially other documents.
The archival description will be analytical, with the aim of examining the data in its entirety, from the folder level down to individual digital documents corresponding to documentary units.
The analysis of the archival collection reveals two timelines: a personal one, corresponding to Cordero's more intimate and intellectual dimension, expressed through his "Diary entries concerning personal experiences, contacts, notes, and reflections of a personal, professional, and political nature", a diary that Cordero kept from June 2002 to June 2017; and a public one, marked by his publications and newspaper articles written for Italy's leading national newspapers.
From a preservation perspective, it will be necessary to migrate files to PDF/A format, suitable for ensuring long-term preservation, while retaining JPG, MP3, and WMV formats, converting emails to .eml format, and converting any attachments to PDF/A.
The AI-powered web platform for integrating the Polo's cultural archives includes two distinct categories of users (administrators and the general public), with differentiated access to the contents of the Cordero archive.
The project includes the definition of access policies and visibility markers for metadata, the implementation of server-side access rules and interface-level filtering controls.
It also involves the development of a user administration page for managing registered users' roles, the adaptation of existing views to comply with the defined access policies, and website optimisation for search engines and content sharing. This includes configuring meta tags, preparing previews for sharing on social networks, and implementing structured data to enhance search engine results.
The AI-based Work Included:
Multi-stage Extraction and Verification:
Every person, place, work, event, and organisation extracted from the archives undergoes a two-stage verification process.
First, an AI model identifies and classifies candidate entities. In the second stage, the model is required to cite the exact passage from the original text that supports each extraction.
This means that every extracted entity is linked to a specific passage in the original text. When users hover over an entity in the sidebar, the corresponding passage is highlighted in the text, allowing them to see exactly what the AI has identified.
If the model cannot identify a specific supporting passage, the entity is discarded.
Locally Deployed Models:
The core of the extraction pipeline runs on self-hosted infrastructure.
These models were selected for their performance and their ability to run entirely on local hardware, ensuring full control over the data-processing environment.
Zero-Data-Retention API Models:
For certain specific and complex tasks, such as disambiguation or cross-referencing information across hundreds of pages, we use models accessed through APIs.
Only non-sensitive data is transmitted, and we exclusively select providers that guarantee zero data retention: prompts and outputs are not stored after the response has been generated.
The commitments from the Polo:
- Full traceability: Every AI-extracted entity is linked to the relevant passage in the source text.
- Data sovereignty: Only non-sensitive data is transmitted to external services.
- Transparency: The AI models used are documented, and their role within the processing pipeline is clearly described.
- Open models: We prioritise open-source and commercially usable AI models that can be adopted by any institution.
Here some screenshots of the platform / archive:
Portugal - APAP at 50: Digitising the Founding Decades of Iberian Landscape Architecture Records with HTR
This project is coordinated by Maria João Fonseca, a landscape architect and researcher at HTC–CFE (NOVA FCSH, Lisbon), working at the intersection of landscape architecture and archival studies since 2019. She is Vice-Chair of the COST Action ConnectLAA (CA23128), Secretary of the ICA Section on Architectural Records, and Principal Investigator of "Reclaiming Their Narrative: The Contributions of Women to Landscape Architecture in Portugal". She works with Anna Laura Jeschke, professor of Landscape Architecture at Universidad Rey Juan Carlos (Madrid), whose research focuses on resilient urban green spaces and enduring open-space design.
Through the ILAM platform (Iberian Landscape Architecture Memory, NOVA FCSH; ilam.pt), and marking APAP's 50th anniversary in 2026, they will digitise and apply Handwritten Text Recognition (HTR) to the founding minute books of the Portuguese (APAP) and Spanish (AEP) landscape architecture associations, with full-text search across both languages.
The records will be described in ILAM, which is built on AtoM (Access to Memory), and ingested into Archives Portal Europe, becoming the first landscape architecture collections available through the portal.
Read the reports of the previous APE Digitisation Grants winners:
- The Military History Collection - Koç University Suna Kıraç Library (Report)
- The Titus Brandsma family archive (Report)
- The Oury Archive (Report)
- The National Archives of Latvia (Report)
- The National Bank of Greece (Report)
