Exploring museum collections online: Some background reading

Published in

Science Museum Group Digital Lab

11 min readJan 23, 2018

In 2017 the Science Museum Group relaunched its online collection website. The website consolidated a number of existing websites publishing digitised collection material — organised by subject or area of collection — into a single presence.

Over 282,000 objects and archival records were published and over 27,000 of these were illustrated with at least one image.

This represents about 6% of the object collection (estimated to total 425,000 objects) and only a tiny fraction of the archival collection (estimated to be around 7 million items).

Beginning in 2018, the Science Museum Group will undertake a large-scale digitisation project. Through this work we hope to digitise approximately 240,000 objects and publish them online.

According to the 2015 ENUMERATE survey of European cultural heritage institutions, science and technology collections are 19.5% imaged and online, so this ambitious project is a step change in digital access to the Science Museum Group Collection and will take us from significantly behind the sector to becoming a sector leader.

As we consider this ambitious project and the impact it will have on the Group’s collection website, we are giving some thought to how a diverse range of audience types will discover and use the collections. Where possible collection images are published under a Creative Commons licence and the catalogue is available via a public API (application programming interface).

However, as is the norm for most cultural collections online, the primary mode of discovery for the content is via search. This limits discovery to those who already know the collection and are looking for something they know is there or speculative searches which may or may not return results depending on what is in the collection, what has been digitised and how it has been catalogued.

Over recent years, a number of cultural institutions, individuals, companies and academic researchers have begun to explore new forms of discovery for cultural heritage collections online which address this limitation and seek to offer new forms of browsing and discovering digitised gallery, library, archive and museum (GLAM) collections.

As we embark on our project we are keen to explore how we might facilitate audience discovery of the wealth of material that will shortly be published online.

This blog post outlines some of our background reading…

“Generous interfaces”

In 2015, Mitchel Whitelaw published an article in Digital Humanities Quarterly called “Generous Interfaces for Digital Cultural Collections”. In the paper Whitelaw argues that:

“[…] search, as the dominant interface to our cultural collections, is inadequate. Keyword search is ungenerous: it demands a query, discourages exploration, and withholds more than it provides. This paper argues instead for generous interfaces that better match both the ethos of collecting institutions, and the opportunities of the contemporary web. Generous interfaces provide rich, navigable representations of large digital collections; they invite exploration and support browsing, using overviews to establish context and maintain orientation while revealing detail at multiple scales.”

Whitelaw has presented his work in the field of generous interfaces several times including at TEDxCanberra (2010) and at the Art Asia Archive (2015).

Mitchell Whitelaw — Visualising culture — TEDxCanberrra

Mitchell Whitelaw — Generous Interfaces: Exploring Cultural Collections — Asia Art Archive

Whitelaw’s key points from a 2014 talk are summarised in a blog post by John Coburn:

Show first, don’t ask. Non-specialist audiences don’t like search boxes. Generous interfaces should encourage easy enquiry.
Provide rich overviews of collections, not entry points. Don’t dictate user journeys. Don’t narrow their experience.
Characterise collections with the collections themselves. How can a visualisation of 7,000 photographs, ordered by the decade they were taken, or the name of their creator, convey patterns and meaning unique to that collection?
Show relationships between items within the collections.
Provide clues. Partially completed and incomplete records are intriguing. They prompt exploration.
Publish rich primary content. Good images. Stables URLs. Without this you have nothing.

Some approaches and examples

Lightboxes

Possibly one of the earliest attempts to explore new forms of online content discovery was San Francisco Museum of Modern Art’s (SFMOMA) ArtScope in 2007 (now decommissioned) which presented the collection in a lightbox-like grid over which a magnifying glass could be dragged to explore individual artworks.

SFMOMA: ArtScope

In 2015, Tyne & Wear Archives & Museums created Collection Dive (working title: Past Paths) funded by the Arts Council/NESTA/AHRC Digital R&D Fund for the Arts. The project provided extensive research and evaluation. The Collection Dive website analyses visitor behaviour to provide “similar” or “not similar” content.

Tyne & Wear Archives & Museums: Collection Dive

As part of his research, Mitchel Whitelaw created Trove Mosaic which presents search results from the National Library of Australia’s Trove website in a lightbox format with a number of viewing parameters including collection grouping and decade.

National Library of Australia: Trove Mosaic

Locations and maps

The Royal Collections website features a Royal Collection Near You web interface which has mapped the artworks in the collection depending on places depicted, places of manufacture, where they can be seen, and places where gifts given to British monarchs originated.

The New York Public Library has undertaken a similar map interface called OldNYC showing historic photographs of the city on an interactive map.

The BFI’s Britain On Film map locates films in their collection on a map and also offers filters by decade and subject, as well as suggested searches.

Colour and visual appearance

A number of collections (notably those with design or fine art collections for obvious reasons) have implemented online collection browse tools based on colour including the Cooper Hewitt, Rijksmuseum, Dallas Museum of Art, and the V&A.

The Barnes Foundation is currently (winter 2017) developing a new web interface to their collection which attempts to honour the “ensemble” spirit of their gallery hang. The project is still a work in progress and was soft launched in October 2017. The team at the Barnes continue to blog about the project and its technical complexity and the use of computer vision to tag the collection and power the interface.

Barnes Foundation: Our Collection (work in progress)

Chronology: Histograms and timelines

A number of projects have added histograms to search results pages to visualise the distribution of results through time. Notable examples are the State Library of Queensland’s Discover the Queenslander website and the Centre for Australian Art’s Australian Prints + Printmaking website.

Centre for Australian Art’s: Australian Prints + Printmaking

Display of heritage collections naturally lends itself to display structured by time. Stephen Boyd Davis (Professor of Design Research at the Royal College of Art) provides a oversight of the history of timelines in the article History on the Line: Time as Dimension (pdf) and his talk Inventing the timeline — a history of visual history.

PhD candidate in Innovation Design Engineering at the Royal College of Art, Olivia Vane has undertaken some exploratory work at the Cooper Hewitt looking at how the collection might be displayed using a combination of timelines and tags.

Cooper Hewitt collection interface by Olivia Vane

The British Museum (in partnership with the Google Cultural Institute) developed The Museum of the World an interactive timeline featuring highlights from the museum’s collection.

British Museum and Google Cultural Institute: The Museum of the World

The Google Arts and Culture Experiments website features a number of user interfaces including the Curator Table that allows search results to be viewed in a timeline.

Google Cultural Institute: Curator Table

The V&A Spelunker: Date Graph displays the entire V&A collection by date and acquisition date (the date the object joined the V&A colleciton).

V&A Spelunker: Date Graph by Good Form and Spectacle

Keyword

Mitchel Whitelaw’s Exploring the Manly Local Studies Image Library interface uses a combination of lightbox and keywords to display results via the National Library of Australia’s Trove API.

State Library of Queensland’s Discover the Queenslander also features a tag cloud as part of the search and browse interface.

The Google Arts and Culture Experiments includes one driven by machine generated keyword tags.

Multi-dimensional / hybrid

Whereas most of the above examples focus on a single “dimension”, some interfaces combine several allowing exploration across multiple criteria. For example, the State Library of Queensland’s Discover the Queenslander website includes a histogram, keywords and colour into a single interface.

State Library of Queensland: Discover the Queenslander

The Urban Complexity Lab at the University of Applied Sciences Potsdam has as one of its research themes Visualizing Cultural Collections (VIKUS). The lab has produced a number of interfaces that allow exploration across multiple dimensions, including:

Past Visions allowing exploration of drawings by Frederick William IV are part of the collection of prints and drawings of the Prussian Palaces and Gardens Foundation Berlin-Brandenburg
Deutsche Digitale Bibliothek Visualized which attempts to visualse the 7 million items in the German Digital Library
Coins allowing organisation and exploration of the coins from the Münzkabinett Berlin collection

Marian Dörk (research professor for information visualization at the Institute for Urban Futures) outlines the group’s approach in a webinar entitled Visualizing Cultural Data: Experimental Interfaces for Digitized Collections from June 2016.

Past visions penned by Frederick William IV by Christopher Pietsch, Katrin Glinka and Marian Dörk

Coins by Flavio Gortana, Franziska von Tenspolde, Daniela Guhlmann and Marian Dörk

The BFI Filmography website allows exploration of a catalogue of UK feature films via numerous dimensions each with its own interface: maps, graphs, charts, etc..

Two Way Street — an independent explorer for The British Museum collection by Good, Form & Spectacle — builds on the museum’s open data initiative and allows exploration of the museum’s acquisitions through a multitude of dimensions including subject, material, technique and date.

Serendipity

One of the joys of visiting a museum is coming across unexpected and previously unknown objects while wandering through the museum’s galleries and displays. A number of initiatives have endevoured to replicate this experience in the digital realm.

The Magic Tate Ball developed for Tate by Thought Den is a mobile app which draws on a number of data sources — time, date, location, weather, ambient noise levels, etc. — and then provides an artwork based on these sources.

Magic Tate Ball

San Francisco Museum of Modern Art (SFMOMA)’s project Send Me SFMOMA uses SMS text messages to reply to messages with artworks from the collection based on the content of the message.

Winner of the 2016 IK Prize was Recognition which used artificial intellegence to analyse photographs in the Reuters news feed and to find an artworks in the Tate collection to pair them with. Images were analysed for objects, faces, composition and context.

Building Recognition

With the growth of APIs for museum collections (ways in which the museum’s collection can be accessed via another computer programme — here’s the Science Museum’s collection API), computer progammers have begun connecting museum collections to social media allowing computer programmes (“bots”) to post images at random or based on some criteria. Dozens of museum bots now exists mostly on Twitter but also on Tumblr and other platforms.

Our challenges and constraints

Our early thinking about development of a new discovery tool or set of tools have thrown up some areas where we think we face some challenges:

The collection is heterogeneous with an incredible diversity of collections — bicycles, medical bottles, locomotives, artworks, computers, musical instruments, etc. — which would need (or possibly not?) to be represented in the discovery interface.
Our digitisation project is rapid and prioritises breadth over depth so it is likely that for the vast majority of objects catalogue data will be thin with only a handful of fields populated at first.
This means that there will be few data dimensions to use, although we’re wondering about using additional data sources such as Wikidata, Dbpedia, image recognition and optical character recognition to gather more data.
Structured data such controlled lists, vocabularies and other ways of interlinking related catalogue records are only sparsely used.
Within the existing catalogue there is some imperfections and gaps in catalogue record data that could well be amplified by a rich, interactive user interface.
Finally, one of the joys of visiting museums is serendipitous discovery of interesting and inspiring objects, how might this be replicated digitally?

Note: Additional examples are continuing to be collected on this Pinterest board.