Research has shown that users have difficulties finding images which illustrate abstract concepts. We carried out a user study that confirms the finding that the selection of search terms is perceived difficult and that users find the subjectivity of abstract concepts problematic. In addition, we found that abstract queries were mostly used to get inspiration for followup queries. Search terms became more concrete as the search task progressed and the idea of the final image took shape. Based on this, we specified three user requirements: (1) representation of multiple interpretations for a single abstract concept, (2) ability to include synonym image results based on the initial query and (3) the need to support the transition of abstract to concrete search queries. We translated the user requirements into an interface design and evaluated the mockup with end-users. All three corresponding functionalities were perceived useful by participants. The ability to select a sub-concept of an abstract concept was reported most useful, because it improves the precision of the returned images.
, , ,
CWI
Information Systems [INS]
Human-Centered Data Analytics

Besseling, R. (2011). Designing an image retrieval interface for abstract concepts within the domain of journalism. Information Systems [INS]. CWI.