Achieving consistency in collaborative image annotation systems

Abstract

Collaborative image annotation is a useful strategy for assigning a set of labels, or keywords to an image, taking into account its content. While existing collaborative image annotation frameworks facilitate sharing, indexing, and understanding of huge number of images, newly-developed methods let group of users manage dynamically-updating data. Conflict-free Replicated Data Type method, used in several collaborative systems, offers to distributed participants through the web a real support to concurrently annotate copies of shared image file regardless of their order. A major benefit of this method, therefore, is to provide a good framework for representing semantic stores and a good mechanism for solving the concurrent updating problem without complex control in collaborative image annotation process. In this paper, we propose a novel optimistic replication approach for collaborative replicated image annotation stores that ensures eventual consistency. We describe how Open Annotation Collaboration can be extended to further support real collaboration between users allowing them to work together effectively to achieve common goals.