A private collection has achieved 50 million image views in one year by bulk-uploading images to Wikimedia Commons and improving relevant Wikipedia articles
by Martin Poulter, Wikimedian In Residence, February 2022
The Khalili Collections are eight collections of cultural treasures amassed by the collector, scholar, and philanthropist Professor Sir David Khalili. They include the world's largest private collection of Islamic Art, a collection of Meiji era Japanese art comparable only to the collection of the Japanese imperial family, a collection of Enamels of the World, and one of the world's largest collections of objects relating to the holy sites of Islam. Sir David has funded the conservation, curation, and exhibition of these collections — which total around 35,000 objects — and their publication in nearly 100 volumes with contributions from leading researchers.
The Collections use many platforms to bring digital images to the public. As well as an official web site and various social media channels, the Khalili Collections has partnerships with Google Arts and Culture and with Europeana, and recently created an interactive online exhibition using Sphere visualisation technology.
Beginning in February 2020, the Khalili Collections employed a part-time Wikimedian In Residence (WIR) to share material on Wikipedia, Commons, and Wikidata. 1,500 images were selected for sharing on Commons under open licences: 1,000 relating to Islam and the rest from the other collections. The WIR works closely with Wikimedia UK, who provide regular advice and support.
In the latest calendar year (2021), the images had more than 50 million views in total: this despite the majority of the images being uploaded in July. As is true of many cultural heritage institutions, Wikipedia articles have become the main way the public encounter the Collections' digital images, far outstripping the views of the Collections' own web site and social media.
As well as this enormous reach, the project has had other kinds of impact:
- Wikipedia volunteers have translated articles about the collections into new languages, including Vietnamese, Malay, and Italian. Image captions have been translated into many more languages, with nearly seventy language versions of Wikipedia now using Khalili Collections images.
- We have added thousands of incoming links to the Khalili Collections web site and hundreds of citations to its paper publications.
- Sharing on Wikidata has opened up new ways of exploring a collection, or exploring across multiple collections. For example, a selection from the Enamels of the World collection can be viewed on a Wikidata map. Qurans in the Khalili Collection of Islamic Art can be viewed in the timeline tool HistropediaJS with links to entries in the online catalogue. When we ask Wikidata for objects related in some way to the Ottoman Sultan Abdülmecid I, we get three objects from different Khalili Collections, plus paintings in other collections (as of February 2022).
- Khalili Collections images are now visible in nearly 900 categories on Wikimedia Commons, alongside related images from other contributing partners such as the Walters Art Museum and the Metropolitan Museum of Art. These categories range over all sorts of topics, from Persian bookbinding to People at work in art; from Golden Horde artefacts to Paintings of Ramayana (Khalili Collections images appear in these alphabetic lists under "K").
Enhancing the reach of images Edit
The 50 million figure for reach includes two broad components, which could be called long-term and short-term. Long-term is when an image is permanently added to Wikipedia and gets a steady rate of views every day, indefinitely into the future. Short-term reach is when an image or article attracts a huge amount of attention in a short time, usually by being showcased on the front page. We are promoting long-term reach in multiple ways:
- Adding images to relevant high-traffic Wikipedia articles
- Finding articles about obscure topics for which Wikipedia lacks an image
- Writing new articles, or article sections, to fill gaps in Wikipedia's coverage
English Wikipedia's front page has five to six million views each day. Other Wikimedia platforms have smaller, but still significant, readerships for their home pages. We used two processes to get images on the front page:
- Putting articles through Did You Know review
- Putting images through Featured Image review
Short-term reach promotes long-term; as well as bringing images to a wider audience, appearances on the front page bring new or high-quality articles to the attention of Wikipedia volunteers, prompting more image uses and article translations, hence more long-term use of the images.
Improving Wikipedia Edit
With some help from Wikipedia volunteers, we have been able to add images to over 250 articles, from high-traffic general articles to very obscure topics. Articles like Alexander the Great, Mecca, or Maria Feodorovna (Dagmar of Denmark) gets tens of thousands of views per month. There are short articles like Ando Jubei, Alfred Morrison, or Ahmed Karahisari that barely reach 100 views per month, but for which the Khalili Collections have relevant images, sometimes providing the article's only image. English Wikipedia has 16,000 articles relating to Islam, so searching through these for articles which can be illustrated by the Khalili Collection of Islamic Art
Wikipedia lacked articles for some important topics that are each relevant to multiple art works in the Collections, so we took the initiative in creating them and getting them reviewed by Wikipedia volunteers. We have created a set of articles related to prominent Japanese artists of the Meiji era and have been building another set of articles relating to the sacred textiles of Islam. The Khalili Collections catalogues are accessibly written by prominent scholars and have contextual essays, so have been ideal sources to base Wikipedia articles on. Volunteers have translated new articles into other languages, at a rate of roughly one translation per new English article. At the start of the project, Wikipedia had no articles about the eight collections themselves, so a first priority was creating those: because of Wikipedia's COI (Conflict Of Interest) rule, these were created as drafts for review by independent volunteers, rather than entered directly into the encyclopaedia.
We have not yet made many articles for individual art works in the Collections. English Wikipedia's notability criterion — requiring substantive coverage in three different sources — means that it is not enough for a work to be extensively documented in the catalogue. There are masterpieces in the Khalili Collections that are notable, but identifying them needs a careful examination of published sources. One Japanese garniture attracted repeated coverage in an art trade magazine and was discussed at length in a scholarly paper, making it clearly notable enough for its own article.
Did You Know? Edit
DYK is a section of English Wikipedia's home page that highlights interesting facts from recently-improved articles. For example "Did You Know that that Plácido Zuloaga trained more than two hundred artists to make damascened artworks?" It is updated every 12 or 24 hours. With each block of eight DYK statements there is one image. If only a tiny proportion of home page visitors click through to read the article, that means thousands of extra views. Articles qualify for DYK by being new, by being expanded five times, or by passing Good Article review. They are each reviewed by an independent volunteer for the quality of their writing and sourcing. During 2020 and 2021 we got 17 articles through DYK. More images are submitted to DYK than can be put on the front page, and an administrator chooses between them. This means there have been occasions when the DYK has been accepted but the image does not make it to the front page.
Featured Images Edit
Images that meet the very highest standards of technical quality, aesthetic value, and educational or historical significance can be awarded Featured status, and Featured images can be showcased as Picture of the Day (POTD). Most language versions of Wikipedia use the POTD selected by Commons, but some Wikipedias — including English, Persian, and Arabic — have their own process for choosing Featured Images and POTDs. After being nominated by the WIR, four images passed Featured Image review on Commons and two on English Wikipedia. The Arabic and Persian Wikipedias have chosen Khalili images for Featured status and POTD without prompting, displaying those images to hundreds of thousands of viewers on their respective front pages.
Lessons learned Edit
It has been hard to predict when things will happen, because of processes like DYK that depend on volunteer reviewers. The quickest DYK reviews have been completed in hours while some take months. It is not possible to predict in advance if a review will be quick. Some images are scheduled to appear on the front page but the date is changed at the last minute for various understandable reasons. So we have adapted to not knowing with certainty when an outcome is going to happen.
Once the images are uploaded, there is a lot of ongoing effort needed to make them easily findable by the Wikimedia community. This takes place both on Wikidata and on Wikimedia Commons, which presently have somewhat redundant ways of linking images to topics. Broadly, on Commons we describe the digital images and on Wikidata we describe the objects. To say that an image is relevant to Aurangzeb, it should be put in the Aurangzeb category on Commons and the Wikidata representation of the object should also be linked to Aurangzeb. Some category tags are added as part of the bulk upload process — for instance the categories for each collection — but the more diversity there is in the collection, the more manual tagging is required. Collections like the Khalili Collection of Islamic Art and Khalili Collection of Enamels of the World contain many different kinds of object, so it is taking ongoing work to properly categorise them. OpenRefine software and the Wikidata/Wikipedia extension for Google Sheets have both been very useful in speeding up the Wikidata process. Volunteer communities have also been very helpful in adding images to Commons categories.
Communications strategy Edit
The Khalili Collections are active in developing their main web site, on social media, and with various content partnerships. These different activities support each other:
- The Google Arts and Culture partnership created super-high-resolution images which were awarded Featured Image status when shared on Wikimedia Commons.
- The WIR searched through academic literature to get additional sources for articles, and thus has discovered positive reviews which are now quoted on the main web site. One paper named a Quran folio as from the "Codex Parisino-petropolitanus" and the catalogue was updated to reflect this new name.
- DYK links appearing on the front page of Wikipedia are an "event" that social media can draw attention to.
- The main site monitors and archives the Collections' media coverage, and this has been extremely useful for sourcing Wikipedia articles.
- The visualisations created in Wikidata have been used to explore new ways to present the Collections on the main web site.
Distinct, complementary roles have been established for the web site (complete and authoritative), social media (time-sensitive and responsive) and Wikimedia (long term, mixing images and text from multiple sources).