Generation of Semantic Clouds based on Linked Data for Efficient Multimedia Semantic Annotations

Abstract

The major drawback of existing semantic annotation methods is that they are not intuitive enough for users to easily resolve semantic ambiguities while associating semantic meaning to a chosen keyword. We have developed a semantic-cloud-based annotation scheme in which users can use semantic clouds as the primary interface for semantic annotation, and choose the most appropriate concept among the candidate semantic clouds. The most critical element of this semantic-cloud-based annotation scheme is the method of generating efficient semantic clouds that make users intuitively recognize candidate concepts to be annotated without having any semantic ambiguity. We propose a semantic cloud generation approach that locates essential points to start searching for relevant concepts in Linked Data and then iteratively analyze potential merges of different semantic data. We focus on reducing the complexity of handling a large amount of Linked Data by providing context sensitive traversal of such data. We demonstrate the quality of semantic clouds generated by the proposed approach with a case study.