Visualising cultural collections: “ show
everything!” (Stamen) One of the key provocations in this work has been Stamen’s call to “show everything”. This work has demonstrated that it’s challenging, but possible, to show everything in a collection using interactive visualisation techniques.
a different angle... But here
I’m going to talk about a different approach to the same challenge - how to make rich representations of big, complex cultural collections.
web pages interactive visualisation rich
interactive engaging data-dense in the browser There’s an interesting space emerging between the standard web experience and the more complex, dense, dynamic world of interactive data vis. The focus of this talk is on creating rich, engaging views of collections in the browser. interesting!
Generous: 1. liberal in giving
or sharing; unselfish: 2. large; abundant; ample This seems to fit with the approach of most cultural and collecting institutions. Their mission is to share, to be “liberal in giving” and unselfish. Right?
Generous: 1. liberal in giving
or sharing; unselfish: 2. large; abundant; ample Collections are also nothing if not large, abundant and ample. Treasure houses of cultural riches! So it seems that generosity, as an attitude, is a good fit for cultural institutions and their collections.
a generous collection demands a
generous interface If collecting institutions aspire to generosity in their mission, then their interfaces should also live up to those values.
Generous unGenerous a continuum: If
some interfaces are generous in spirit, then some might also be less so. I am going to argue that many of the conventions of collection interfaces are less than generous...
<ul><li>attitude: “yes, what?” </li></ul><ul><li>provides no
context </li></ul><ul><li>assumes a query </li></ul>Why Search is Stingy (used alone) ^ Search conveys an attitude: “yes, what would you like”? rather than, “here’s our collection!”. It provides no context to the query - how can we search, if we don’t know what’s in the collection? And it assumes that the user has a focused query that can be expressed as a string. A big assumption!
“ since we know ...
that people have substantial difficulties in specifying what it is ... that would help them to realise their goals, only considering specified search as the basis for IR models and techniques, is clearly inadequate, and inappropriate ” Nicholas Belkin, “Some (what) Grand Challenges for Information Retrieval” (2008) ( http://www.asiaa.sinica.edu.tw/~ccchiang/GILIS/LIS/p47-belkin.pdf ) Even “information retrieval” people recognise that search can’t do it alone.
search + samples = instant
context This is work by Tim Sherratt - adding a box of random samples from the collection images, provides a rich, evocative impression of the diversity of the collection. A small (but useful) sense of what’s in there provides useful context for a search.
This LOC example provides samples
of content from “100 years ago today” - again effectively a random core sample of the collection, but far more generous than a blank search box, and immediately evocative.
This search on the Australian
War Memorial collection provides informative facets - but only after the search! Until I search, I have no idea that the collection includes, Art, Film, Heraldry etc.; but these facets reveal collection structure as well as distribution.
These are “browse” facets from
the Art Gallery of New South Wales site. They combine sample images with collection structure and item counts - rich but compact cues for exploring the collection.
A new Generosity? Here, I
want to look at some more radical examples of generous interfaces - and there seem to be more and more appearing. NB this is not a comprehensive list (feel free to suggest good examples).
<ul><li>Mildenhall and AWW </li></ul><ul><li>immersive: “wall
of thumbnails” </li></ul><ul><li>immediate: link to rich content </li></ul><ul><li>exploratory </li></ul>These two projects share an approach that focuses on immersing the audience in evocative but structured collection material. In both cases the interface links to rich content, and it doesn’t assume a query, but encourages exploration.
<ul><li>linked content (mashup) </li></ul><ul><li>samples: partial
information invites exploration </li></ul><ul><li>implicit relationships </li></ul>History Wall Tim Sherratt’s History Wall takes a slightly different approach: “wall of thumbnails” again, but here combined from multiple sources, to create implicit relationships between items. Note that providing samples or excerpts really invites exploration - we want to complete the story, join the pieces.
The decade browser gives an
overview of the whole Prints and Printmaking collection (over 20,000 works) ordered by decade. The decade display is just a histogram - taller towers mean more works. Inside each column, images of the works from that decade are overlaid to give a sense of the “tetxure” of that period. This is built in plain HTML and jQuery, accessing the data through an API developed for us by Tim Sherratt.
Selecting a column will load
a horizontal-scrolling list of all the works in that decade. We can select a work image to link to its full page on the NGA site. Artist names link to the artist explorer interface that we will introduce in a couple of slides.
<ul><li>quantity + visual sampling </li></ul><ul><li>the
“texture” of the times? </li></ul><ul><li>micro in macro context </li></ul>Decade Browser This tool combines a visual representation of quantity (a histogram) with the unstructured, evocative visual samples of the works themselves. We can move from macro overview to micro - item level display - and situate the items in their broader context.
This “artist explorer” started life
as a way for us to get to grips with relationships in the collection. It’s turned into a useful interface, focusing on artists’ works and their collaborations with other artists. In the left hand pane we see an artist’s details and a list of their collaborators. The central pane shows all the artist’s works.
The right hand column shows
a larger view of a single work. Again we are showing micro (full item details) in its macro context (all the artist’s works, and collaborative relationships).
Hovering over the related artists
highlights their linked works - here we can see Aleks Danko’s posters made in collaboration with Colin Little. Note that hovering over works also highlights the artist names - so these facets are two-way linked to the collection items. By clicking a related artist we can explore their work - and so on, spidering across the collection, discovering connections.
<ul><li>linked facets </li></ul><ul><li>micro & macro
in context </li></ul><ul><li>exploring relationships from the inside </li></ul>Artist Explorer Rather than a “bird’s eye” view of a collection, this tool provides an exploratory view from the inside. We can develop a sense of the collection, and relationships within it, by wandering around within it. Linking between overview and immersive modes seems like a promising way to combine different interfaces.
<ul><li>Show first, don’t ask </li></ul><ul><li>(provide
rich information up front) </li></ul>Search is a way of asking first - requiring the user to make the first move, information-wise. Offering more up front is the generous way to do it; it provides context that enriches our understanding of the collection.
<ul><li>Free samples </li></ul>The best way
to represent or characterise a collection is through the items themselves - samples show diversity and distribution; they are evocative and enticing; and they enrich our sense of context.
<ul><li>Share the Good Stuff </li></ul><ul><li>(high
quality primary content) </li></ul>The bottom line of generous digital collections is sharing rich, high quality digital content - and preferably doing it in-situ, without jumping through hoops or switching contexts.
Technical Challenges <ul><li>More data, more
images, more bandwidth </li></ul><ul><li>Need efficient data access (API) </li></ul><ul><li>Requires a modern browser </li></ul>There are some technical challenges to confront if we are going to make really rich, generous interfaces. APIs are essential, but current APIs are designed around search, and drip-feeding individual records, rather than delivering rich samples or overviews of large collections. Delivering it in HTML and JS requires the user to have a modern browser.
Design Challenges <ul><li>Hybrid conventions (web
/ visualisation) </li></ul><ul><li>Unfamiliar interfaces </li></ul><ul><li>How much is too much? </li></ul>This space is rich with challenges for designers / developers. The last question bears some more exploration....
low density high density We
could also frame it as a continuum of data-density - with the conventional web being relatively sparse, and visualisation being more dense.
stingy overwhelming Or we could
think of the continuum as between stingy - too little information - and overwhelming: too much. Now “generous” is actually the sweet spot between these poles. If it’s possible to be too generous, how do we know where the optimum is? generous ?
Thanks! Mitchell Whitelaw Faculty of
Arts and Design University of Canberra [email_address] @mtchl Thanks to the NDF for inviting me to present this work - feedback welcome, of course.