A multi-criteria Recommender System for Tag Annotation

Marc Gräßle, Dirk Werth, Peter Loos

In: Bebo White, Pedro Isaías (Hrsg.). Proceedings of the IADIS International Conference WWW/Internet 2013. IADIS International Conference WWW/Internet (WWW/I-2013) October 22-25 Fort Worth TX United States IADIS 2013.


Web 2.0 signifies an altered use of the internet. Interaction and collaboration are standing in the spotlight. An important role in most Web 2.0 applications is ascribed to the tagging of user generated content. However, the simplicity of tagging entails some limitations. Since every user associates different things with an object, from an objective point of view, tags are often imprecise. We will present an approach that pursues the idea to disseminate a consistent vocabulary amongst all users by providing the users proactively suitable recommendations to tag images. In addition to existing tag-relation based approaches, we include the image analyzes of the image to get graphical and theatrical similar images and then extend the tag recommendation with the retrieved context information.

Deutsches Forschungszentrum für Künstliche Intelligenz
German Research Center for Artificial Intelligence