Personalization in crowd-driven annotation for cultural heritage collections

Many cultural heritage institutions are confronted with a big challenge when it comes to adapting the process of registration, annotation and digitization of their collections to meet the new technological demands for providing their collections online with Web and mobile technologies. With limited funding and limited professional resources, annotation is often lagging behind significantly in terms of quantity and level of detail. Quantity as well as quality of the annotations is the key to finding relevant objects by various groups of users in online collections. As many experiments with cultural heritage data online show, human computation can be useful for collecting large quantities of data. Various of these experimented with a crowd of lay people to help with the annotation of cultural heritage artefacts. However, a common concern relates to the resulting quality. The current trend appears to focus on supporting ‘the crowd’ to achieve the desired level of quality. In this paper, we discuss the results of the first phase of an evolutionary system design performed in collaboration with the Rijksmuseum Amsterdam, as part of the SEALINCMedia project. By providing personalized support to the individuals and groups in the ‘crowd’ when annotating museum collection objects, we also support the integration of the annotation result in the RMA content management system. The contribution of this paper is twofold: (1) requirements specification for a crowdsourcing annotation process and (2) design of the personalization functionality in a crowdsourcing annotation system.