Thanks for the mention. For the astrolabes work, we are combining data from disparate museum catalogues with data extracted from a draft of an academic publication and sometimes even the metadata associated with individual images. There are elements that don’t-quite-match so it’s usually far from an automatic process.

http://glam-discovery.bodleian.ox.ac.uk/astrolabes/

Short video explaining the project:

https://www.youtube.com/watch?v=95FmWIQiPIw

The interface isn’t yet as good as it could be, but the data set is the higher priority: the most usable ways to visualise the data can come later.

 

More relevant to the Kew case is this application which visualizes an 18^th century botanic expedition, combining data about the 1,900 artworks created with tree-of-life data connecting the species, and geographical data about locations visited.

http://glam-discovery.bodleian.ox.ac.uk/botany/

 

I’m also working on this application which makes connections between objects from some Oxford collections: the Ashmolean, the Bodleian, the Pitt Rivers, by generating profile pages for depicted entities, creators, and locations. OpenRefine and the Wikidata tools built into Google Sheets have been useful for this.

http://glam-discovery.bodleian.ox.ac.uk/

 

Making connections, by drawing together identifiers from different systems, is very much what this work is about. There are some schematic examples of Wikidata connecting across different collections in this blog-post:

http://blogs.bodleian.ox.ac.uk/digital/2019/01/23/making-wikidata-visible/

 

Hope this is of interest,

 

 

 

From: Museums Computer Group [mailto:[log in to unmask]] On Behalf Of Jamie Unwin
Sent: 01 February 2019 13:57
To: [log in to unmask]
Subject: Re: [MCG] Looking for examples of connecting objects/records across collections/institutions

 

James,

 

I’m guessing you’ve already come across Martin Poulter’s Wikidata backed Astrolabe Explorer? Martin’s also doing some interesting stuff across the various Bodleian collections using Wikidata Q codes, might be worth dropping him an email.

http://glam-discovery.bodleian.ox.ac.uk/astrolabes/

 

Jamie

 

From: Museums Computer Group <[log in to unmask]> on behalf of James Morley <[log in to unmask]>
Reply-To: Museums Computer Group <[log in to unmask]>
Date: Friday, 1 February 2019 at 13:42
To: "[log in to unmask]" <[log in to unmask]>
Subject: Looking for examples of connecting objects/records across collections/institutions

 

Hi all

 

I am currently working on a project for Kew Gardens called The Mobile Museum, looking at ways that the Kew collections of ethnographic artifacts and plant raw materials were distributed across the world from around 1850 onwards. These went to other museums, research institutions, commercial companies and, mainly within the UK, schools.

 

We're trying, both manually and digitally, to trace these and connect up any records. The manual part is being handled by researchers and has seen them visit institutions in the UK but also Australia and the US. For my part I am looking at the digital side and starting to explore ways that we can make or at least predict / narrow down those potential connections. This might involve being sent datasets, or accessing them directly (in the few cases where they have APIs) or through services like Europeana and Trove.

 

More details below, but my question is whether anyone has seen any similar projects comparing datasets and trying to link objects between collections? 

 

To show the sorts of things I'm exploring

- we have our own collection of things that were kept (often what was sent would have been duplicates/spares) and this includes metadata of scientific and common names, descriptions, sometimes origins and collectors etc

- we have an 'Exit Book' which has records of what was sent, including where to and when, which also has details of scientific and common plant names; these have been transcribed and enriched with broadly standardised metadata including dates, plant names, recipient institutions, and geography

- in some cases we are in touch with recipient organisations (like the British Museum) and they have provided simple csv record extracts of objects known to have been received from Kew. That's fine for selected major institutions, but there are about 1,000 distinct recipients and 40,000 objects so that's not going to scale, plus it relies on high quality historic record keeping and metadata to even find the data

 

Here's an example:

In 1866 Kew sent some material to the British Museum, and within this was an Iban skirt from Sarawak (indeed before this there are Kew archival records that show it was sent to Kew by James Brooke, the first Rajah of Sarawak). That item was then actually passed on to the Pitt Rivers. We have accession data (transcribed but not yet publicly available) that states “5 pieces of native cloth from Borneo” were received at Kew from Brooke on 24 June 1856, plus details of the geographic origin etc. Through painstaking manual research these have been connected to an item in the Pitt Rivers (available online, but they don't appear to have permalinks to object records) which includes mentions of Sarawak, Iban, and part of the text description reads "The width suggests that it might be a skirt length" plus includes the date 24 June 1854 (which is itself only a partial match as at some point it appears to have been mis-transcribed!). Not a huge amount to go on, but it feels like there could be enough tantalising details to start making connections.

 

I know it's a long shot but if anyone has any ideas and examples of comparing very variable datasets and predicting matches based on metadata and/or textual descriptions (or even visual comparisons, but that's an even longer shot practically and technically!) then I'd love to hear of them.

 

Thanks, 

 

James

 

PS if anyone is interested in this field then we have a conference "Collections in Circulation" happening at Kew, 9-10 May (a wonderful time to visit the gardens!) - see https://www.eventbrite.co.uk/e/collections-in-circulation-international-conference-registration-53049015032

 

 

---
James Morley

Projects: www.catchingtherain.com 

Twitter: @jamesinealing / @PhotosOfThePast

 

 

 

--

Dr Martin Poulter

Wikimedian In Residence at the University of Oxford, based at the Bodleian Libraries

https://www.wikidata.org/wiki/Wikidata:GLAM/Oxford