This paper discusses the complex process of reconciliation of data coming from research and public cultural heritage institutions with their own selection criteria that have shaped the provenance of their collections. We demonstrate how the Golden Agents digital humanities research infrastructure [1] can play an intersecting role as intermediary data provider between these distributed collections in the reconciliation, disambiguation and deduplication of data by taking their provenance into account. To this end we analyse an art historical case we call ‘The Montias Case’, with data from three different sources: the Getty Provenance Index, [2] the Frick Collection, [3] and the notarial acts of the Amsterdam City Archives. [4]
In the 1980s, John Michael Montias (1928-2005) began to compile a database containing records of ownership of works of art of the Dutch Golden Age. To this end he also selected records from the Amsterdam City Archives using inventories dated 1597-1681. In 1985 and 1987, Montias was invited to the Getty Research Institute to automate his work and incorporate it into the Getty Provenance Index. When his collaboration with the Getty concluded around 1990, Montias continued his project and donated his work to the Frick Art Reference Library. Both datasets were enriched by others. The Getty Provenance Index and the Frick Collection have now two partially overlapping datasets with a shared provenance that are in need of deduplication/disambiguation to fully exploit them for art-historical research.
The Getty Research Institute and the Golden Agents consortium set up an experiment in 2019 to reconcile the Montias data with their provenance and enrichments to their original source: the notarial acts. This reconciliation via the Golden Agents infrastructure is necessary because the Amsterdam City Archives due to their public mission only index person and geographical names, but not the art objects that Montias listed that we need as researchers.
We first link the inventories from the Getty Provenance Index and the Frick Collection to their corresponding notarial deeds. Additionally, we deduplicate the names indexed in these data sources and identify people utilizing the Lenticular Lenses II tool (Idrissou et al., 2018, 2019) in development within the Golden Agents project. Then we link the individual items indexed in the Montias database to the transcription of the historical archival source. The Web Annotation Model is used as a data model for the text and serves both modelling and presentational purposes. This allows us to link and to validate the exact piece of text indicated by coordinates on the scan where a certain object is listed and to present it for instance in a IIIF-environment. The thesauri AAT and TGN, [5] and ICONCLASS [6] help in describing these items in a standardized way. Finally, we demonstrate how, in line with the goals of the Golden Agents and the remodel of the Getty Provenance Index as Linked Open Data projects, [7] the results of the Montias Case can be relevant for data reconciliation and for modeling the provenance of distributed heterogeneous cultural heritage collections for (re-)use in digital humanities research projects in general.
[1] https://www.goldenagents.org
[2] https://www.getty.edu/research/tools/provenance/search.html
[3] https://research.frick.org/montias/
[4] http://archief.amsterdam/archief/5075/
[5] https://www.getty.edu/research/tools/vocabularies/
[6] http://www.iconclass.org/
[7] https://www.getty.edu/research/tools/provenance/provenance_remodel/index.html
References
Bibliography
Web