Art Maps: Mapping an art collection

The new online collection – launched in April 2012 as part of the relaunch of the Tate website and renamed Art & artists as a result of lab-based user testing – endeavours to provide a few such new ways of discovering art. Whereas the previous collection was based largely on the premiss that the user already knew what they were looking for and that that was a particular artwork, the new collection includes you-may-also-like features prevalent on shopping websites. Users can now move through the entire collection without having to visit the search or browse features, instead jumping from one artwork to another. In addition, the search box now returns results from multiple database fields (not just artwork title) and results can be viewed in multiple ways designed to aid browsing large numbers of artworks: as a list, as a grid of images and as a fullscreen slideshow.

Given the number of artworks in the Art & artists database – over seventy thousand – our plans to create these new interfaces needed to leverage the existing data generated by the digitisation of the collection and the subsequent ongoing acquisitions processes. One area that drew our attention was the ‘places’ branch of the subject index. Over twenty-three thousand artworks have been indexed with one or more places, although this does include some mythical places. Given that maps are such compelling and well understood reference tools for all kinds of information, we began to wonder if the real places could be displayed on a map and to discuss the interesting possibilities that might emerge from this. Our initial thoughts included:

• results from a search for ‘castle’ could be viewed not only as a list, image grid and slideshow but also on a map;
• a J.M.W. Turner sketchbook, such as the Isle of Wight Sketchbook (1795), could be viewed as a map with pins showing the places depicted;
• a location-aware and internet-enabled device such as an iPhone could display artworks depicting or associated with the user’s current location.

Since a number of works could be indexed with the same subject index term, we began by mapping the index terms so that all the works indexed with a given term, such as the forty-seven works indexed Stonehenge would be located in one go. This first phase saw the bottom terms on the place names index (the subject index is a three-level-deep hierarchy) were ‘geocoded’ against a number of available databases to return latitude and longitude information. Among these databases were GeoNames and Google Places. While this captured the vast majority of the places, others were not possible to locate in this way, and futhermore, the locations resulting from this work were limited by the subject indexing undertaken by Tate. Clearly, for the most part, detailed knowledge of the places with which artworks are associated exist not within Tate but within local communities and those with specialist interest in the artworks or the places. There then followed a first crowdsourcing initiative which aimed to locate the remaining terms by drawing on this knowledge. An archive of this can be found in the National Archive’s archive of the Tate website. The mapping of subject index terms proved problematic as the link between artworks and locations was not evident and so the outcomes of the work that users were undertaking remained opaque and uptake was therefore limited. These index terms were subsequently mapped by hand. Thanks for this phase are owed to John Sidney-Woollett who programmed the initial transfer of place index terms to locations and Roy Patterson who subsequently undertook researching and locating the problematic index terms.

 The resulting dataset from this phase is the foundation for the Art Maps project. Art Maps works towards realising our ambitions and will see our online audiences map individual artworks, clearly something that is far more compelling than mapping the most challenging places terms in our subject index. As an experiment I used the iPhone EveryTrail app to build a walking tour through Winchester linking the places which are depicted in works in the Tate collection. These locations were relatively well known and relatively simple to locate. However, mapping individual artworks throws up some interesting challengers including:

• Artworks may depict a view with a number of landmarks shown, such J.M.W. Turner’s London from Greenwich Park (exhibited 1809). Should the landmarks be mapped or the hill? Are these the same kind of markers or different?
• How might a generic scene be handled, such as a Maria Helena Vieira da Silva’s Paris (1951)?
• How might we map a single sketch book page with multiple sketches on it, such as J.M.W. Turner’s Four Sketches at L’Ariccia (1819)?
• How might we map regions such as a country, for example to map Hans Haacke’s work A Breed Apart (1978) which is related to the South Africa of the apartheid era?
• How might we map a route such as Hamish Fulton’s work on The Pilgrim’s Way (1971)?
• How might we map the locations of the materials that are the media of a work such as Henry Moore’s use of Hornton stone?
• How might we map the places associated with the performance of a work or it’s installation, for example, Mike Nelson’s Coral Reef (2000) and Matt’s Gallery?
• How might we map an artwork that is itself a map, such as Kathy Prendergast’s Lost 1999?

In discussion with Rebecca Sinker and Emily Pringle from the Tate Learning department, new questions emerged that reflected their interest in how the very act of engaging with artworks in this way might not only provide valuable data to Tate, but also deliver the fundamentals of the Tate Learning strategy and thus Tate’s mission itself. Their involvement has been key to the shaping the Art Maps project and resulted in new activities emerging to look at new questions such as:

• How might we map more personal associations between artwork and place such as ’reminds me of’?
• How might the ability for audiences to upload their own content shape their relationship with the act of mapping artworks?

The coming months will see exciting technical developments and trails with user groups which will address many of the questions above and which we shall continue to document on this blog.