After some time, I got involved with a hack on some museum data ina group that included Lisa Gee and Sophie Dixon. I had hoped to do something towards the research and I got more out of it than I had hoped. As my interest is more in the area of how data is constructed and middleware, I got more into the behind the scenes parts of the museum.
Talking with the data owner, Ananda Rutherford, and as part of a series of hacks, I began looking at the way that a museum artefact is represented in the database against how it appears on the website. Through the Geffrye museum website, an artefact can be found and links for either the collection or linked objects. It appeared that the museum has a simple API (Application Programming Interface), which was not as well known as it might be.
Following a suggestion, I focused my attentions on the “Documenting Homes” collection and the data objects inside it. Having downloaded the data, I extracted various objects and wrote them out.
I found that this provided two ways of describing the object in question. The initial one is the curatorial description, describing the details of the item from a curation perspective and how this is to be used on the website. It is the most prominent one.
There are also web keywords set up as a taxonomy. This allows the public to search for items linked to that keyword. As far as I am aware, this is not entirely fixed in stone. It can be updated through the database but it is not folksonomical. Though far more formal than the curatorial description, it reveals how the object may be thought of internally. As a thought experiment, it might be revealing to the use this from the website to see what comes up and how it might be used to redirect the user. Perhaps it might be a serendipitous journey to find other linked objects and to view how the museum collections may be alternatively used.
The initial data dive showed how one object has two ways of being described: either a free text field to add in expertise or how it is tagged as part of the website infrastructure. If I follow the work of Bowker and Starr (2000) and their argument about infrastructure hiding in plain sight and use, then this work begins to surface these descriptions. Yet it also falls back into hiding infrastructure. The collection description is part of the narrative, it is describes the object as part of a collection. However the object is being described in other ways and does not necessarily require the collection tag. Perhaps this raises a question about collections as infrastructure and knowledge representation.
Whilst extracting the data, I came across the materials field. This describes what materials used, such as clay or watercolour, and they are intimate to the object, describing its fabric. The object is likely to be made of something though this might be digitally born or physically made up.
Writing this post has made me realise how all of these models are constrained and brought into being by a fifth model: the database itself. The models might describe and object but they are intact constrained and described the database types and fields. That way opens up some intriguing possibilities of how the interactions affect what we view.
The data is creating a narrative through its digital mediation and transformations from the database to the screen. Equally it is created through a series of narratives encoded as fields and their types, echoing how the museum sees the artefact.
I have put all of my scripts up into Github: https://github.com/iaine/collectionsdata. In due course, I hope to be able to continue developing the code.
Bowker, GC and Star S L. 2000 Sorting Things Out: Classification and its Consequences. MIT Press, Cambridge, Mass.