Towards A Collaborative Framework for Image Annotation and Search

Users tag images with plain text information, which is then used as the basis for search. For the large amount of digital images available on the web this becomes challenging because the tags are abstract concepts whose relationship is undefined. For effective search which requires reasoning on concepts and their relations one requires richer data structures for tagging and one needs to take into account the confidence and credibility of the tagging user. In this paper, we introduce a novel collaborative framework for image annotation, which allows users to create tags that are based on a concept repository which provides a hierarchical context for them as well as allowing to define relationships among said concepts. It also provides a new and systematic way to establish user credibility as well as to compute the truthfulness or reliability of a particular statement, which are used for ranking search results. A prototype has been implemented using this approach and we will show some examples to explain our methodology in detail.