Personalized Tag Recommendations to Enhance User´s Perception

Tagging is a process whereby users freely choose keywords to label Web objects in order to share or recover them later. Tags associated to an object by the user depict his viewpoint or perception. The perception of the target user can be enhanced by aggregating and analyzing the tags associated to an object by other users who share at least one tag with the target user in common for that object and recommending tags to him. Most of the existing techniques recommend tags on the basis of their popularity among the users. The proposed approach complement these approaches by taking the temporal nature of interests of the user into account and enhancing his perception by analyzing and finding relationship among the tags associated to an object by him and other users. This relationship is analyzed using pheromone updating strategy known from ant algorithms for computing the weights on the edges of the co-occurrence graph containing tags as nodes. To observe the performance of our approach, experiments are carried out on the data collected from delicious, a social book marking site that allows the users to tag URLs and share them with other people.