Using controlled vocabularies to organise digital images for improved search results

Click the button below to download the full text of the article.

Abstract: Controlled vocabularies are essential tools for enabling search within a collection of assets. When constructing a controlled vocabulary for a collection of digital images, however, it is important to consider the visual nature of images. The most important principle when designing and using a controlled vocabulary to organise an image collection is that a term should be associated with an image if and only if a user searching the collection would expect and want the image to appear in the result set when searching for that term. This paper explores taxonomy development and maintenance for digital image collections, how hierarchical standards might vary between image and text-based collections, and the impact of keywords on precision and recall.

Keywords: taxonomy; controlled vocabulary; search; image retrieval; image indexing

Ann Pool graduated from the University of Washington in 2013 with a master’s degree in library and information science. She has worked at Corbis as a search metadata technical specialist, focusing on using crowdsourcing campaigns to improve the quality of search keywords associated with digital images.

Laura Horan has been a search vocabulary specialist at Corbis, where she worked to improve search through taxonomy, metadata improvements and communication with contributors. She graduated with a master’s degree in library and information science from the University of Washington in 2014.