Semantic Pattern-based Automatic Annotation Process of Images Shared on Social Networks

Abstract:

Nowadays multimedia document sharing networks, in particular images, have become very popular. they enables to host a massive amount of data stocked in the databases. Therefore, the discovery of the most relevant images, Which respond  to the users needs, constitutes a challenge for researches. Thus, many research studies have been focusing on improving the images annotation process.In fact an effective annotation helps to facilitate and optimize the images research process. In this context, we suggest in this paper an annotation approach dedicated to the socio-tagged images. The suggested approach consists of constructing and inferring a set of semantic rules that specifies the different semantics of an image according its semantic facets pattern. These rules are presented via the logic of predicates experimental studies studies are performed on a collection of 25.000 socio-tagged images shared in Flickr service and the evaluation results demonstrate the effectiveness of the suggested approach.