Today we launch our team effort to propose some new heritage fields for IPTC. The background to the project is in my previous blog posts, but the long and short of it is that images set loose from their databases and web site need data embedded in them, both for general attribution purposes - increasingly important as museums increasinglymake images available to the public for download and use - and also for transfer of data and general information.
This week I stood in in front of two Van Gogh sunflower paintings, one loaned from the Van Gogh Museum in Amsterdam, and the other from the National Gallery collection. The paintings are very similar, though not identical. I downloaded the images below from the National Gallery web site and was pleased to see that there was enough information embedded in each of the images to identify the museum holding the painting, though the National Gallery had no information about the painting or artist, and neither museum used the IPTC Extension artwork fields. But this is progress of a sort; I have spent many years looking at unidentified Van Gogh Sunflowers to demonstrate the point about attribution. It matters!
The metadata from each image is visible at the bottom of this post, as I want now to get onto the forward looking stuff.
We have a group of around 20 professionals from galleries and museums in UK, Europe and USA helping us identify which fields we should add to the IPTC Extension schema under the 'Artwork or Object in the image ' section. We will then go further and identify fields for use in an extended schema we are calling SCREM, which will be useful for data workflow and exchange within and between organisations. This will consist as far as possible of fields from existing well used schemas.
The group has scrutinised our general approach and there is broad agreement that the IPTC fields are for attribution and general display to the public, while the SCREM fields will cater for more in depth information needed by organisations.
A number of issues arose in the course of our consultation, which are worth mentioning.
1. Authoritative Data
Nothing we do with embedded data alters the fact that the collections management database of a museum or gallery is the authoritative source of data; this is where the dynamic and curated data about the object is stored. Embedded data is useful as a snapshot or viewing version of the data, needed for data transfer, or for simply viewing what is known about an object (like, for example, a label on a painting hanging in a museum). The internet is now one vast museum site, but there is an advantage we didn't have before when we stuck paper labels onto prints. Now we can make a link to a url or uri on the internet where the dynamic, authoritative data is kept. That is one of our proposed fields, and it is key to understanding why embedded data and database data are not in opposition, but cam work together very effectively.
Some people argue that embedded data is not useful because it is not persistent. It can of course be stripped by software, though it is mostly by oversight than intent. But the usefulness of embedded data is such that it is better to have it and use it, and persuade software providers to retain it, than to live in world where images swirl around without any data at all. This is the 'done not perfect' philosophy by which most of us lead our lives.
3. Human readable or machine readable?
Metadata is used by two very different constituencies; machines which link data to other data (linked data for example, or a uri in embedded data which links to data on a web site) and humans who want to read data and be informed by it. Some people may argue that strings of machine readable data are what we need. so that a link is always made to the authoritative data. My argument will be for both - human beings do not want to click through several links of unintelligable code to find the 'real' data, like the title of the painting, or the artist name. So we should provide access to both, so that key display data is quickly available to read by humans inspecting the image of an artwork, and machine readable data is there as well for resolution at another location.
4. Objects beyond artworks
It was pointed out to us that there are objects in museums that are not artworks, like archeological finds, where information on find spot and collector are very important to understanding the object. We are consulting on this and have included some extra fields in our IPTC candidate list.
If you are working with heritage metadata and want to be involved in the detailed work of looking at data fields, please send an email to either Sarah Saunders firstname.lastname@example.org or Greg Reser email@example.com, and tell us where you work and what your role is. We can add you to our Basecamp project group. Or if you just want to be kept in touch, let us know.