Birdwatchers in the Rijksmuseum demonstrated a new semantic annotation system that invites web users to enter information on the drawing or painting they see.
As a publicity stunt, the presentation was well conceived. The Rijksmuseum, Naturalis and Wikimedia had invited 45 people with knowledge of birds for some bird watching in the museum on World Animal Day (October 4th). After a short instruction, they were shown pictures of birds from scanned drawings or paintings. Their task was to enter the common name of the species they saw, together with additional information and the degree of certainty (‘are you sure?’).
Public television was there, as well as a couple of newspapers to watch the birdwatchers at work. Fascinated by the images and the people classifying them, most media missed the ingenuity of the software called Accurator of which this happening was the launch. What the broadcast referred to as a computer ‘programmaatje’ was the result of six academics working on an innovative information technology project for four years.
"We're still busy analysing the results of last Sunday", said Dr Lora Aroyo over the telephone from the United States. She is the head of Web & Media group at the department of computer science VU Amsterdam, "It was our first large-scale annotation event. I hope it will give us the information we need to improve the Accurator and make it ready for wider use on the internet."
Accurator is designed to recruit wisdom of the crowd to create annotations for museum objects and images. These annotations make the objects in online museum collections better accessible since you can search for them by simply entering the name. Not only do curators not have the time to supply the information at a rate of 40 thousand objects per year (that is the rate at which the Rijksmuseum puts objects online). Also, curators often lack specific expertise on for example birds, plants and animals species. The same applies to Japanese prints (from the Rijksmuseum) or Bible prints (related to Bibles in the Koninklijke Bibliotheek collection). Amateurs, gathering in online communities, often do have the specific knowledge to tag the images correctly. Accurator aims to tap that knowledge to improve the accessibility of our cultural heritage and the information surrounding it.
Anyone can join Accurator, but contributing is another matter. The system currently invites expertise either on birds or bibles. If the user doesn't claim any specific bird knowledge, she is presented with random objects featuring birds and invited to fill in the genus, scientific name and other details about the bird.
A smart system monitors the input from various users and judges the reliability of the information by comparing the input with that from others and keeping track records of the contributors.
Accurator is the outcome of the SEALINCMedia project (socially enriched access to linked cultural media one of sixteen research projects of the public-private COMMIT/ research community dedicated to solving grand challenges in information and communication science. Researchers from TU Delft led by prof. Geert-Jan Houben cooperated with VU University and CWI in Amsterdam. Cultural partners were Rijksmuseum, Naturalis and Koninklijke Bibliotheek.
Four PhD students worked on various aspects of Accurator:
- Strategies and software to identify potential contributors by Jasper Oosterman, PhD student with prof. Geert-Jan Houben (Web Information Systems) at the EEMCS Faculty.
- Design, development and evaluation of personalized semantic search strategies by Chris Dijkshoorn, PhD student with Prof. Guus Schreiber and dr. Lora Aroyo (VU Amsterdam) at the Computer Science Department
- Management of trust and authority for accessing, integrating and distributing information resources by Achana Nottamkandath, PhD (VU Amsterdam)
- Design, development and evaluation of user interfaces for searching linked data by Mieke Leysen and Myriam Traub (CWI – Centre for Mathematics and Informatics, Amsterdam).
Read also: Expertise from the crowd, Delta, February 04, 2014