User Adaptive Photo Tag Recommendation Employing Social Networks

Sharing photos via social media sites such as Facebook, Picasa, Flickr has become a very popular way to manage and exchange personal photos in recent years. From the vast pool of photos, retrieving pictures is facilitated by the usage of tags/annotations. These tags can either be user-defined or system-recommended. In this work, we propose an approach to recommend tags on Flickr, for partially annotated or un-annotated images, with the help of the data from the user’s existing images and their associated tags, social relations and images from the Flicker database. We propose a method to boost the occurrence of some important tags based on a tag augmentation scheme utilizing the social interactions of the users.