Strategy Evaluation for High Quality Crowd Annotations in Cultural Heritage

Abstract

Digitised Cultural Heritage Objects like pictures, video and audio are hard to retrieve without metadata like annotations describing what is depicted or the events which take place. Unfortunately the creation of this metadata is quite a slow and lengthy process, especially when this needs to be done in a qualitative manner. So the question arises how we can help to improve this process by increasing the speed and quantity of these annotations without sacrificing the quality. The chosen approach is to use lay people and ask them to help us with annotating these objects. As a means to guard our quality demands we want to use a taxonomy which allows the lay people to annotate in a structured manner. In order to help lay people we’ve created a workbench which allows us to easily create different annotation strategies, which result in different ways a taxonomy is presented to a lay person. By presenting different strategies we can investigate which strategy is more likely to help a group of lay persons when this is executed on a larger scale.