A fast effective multi-channeled tag recommender
Jonathan F. Gemmell, Maryam Ramezani, Thomas Schimoler, Laura Christiansen, Bamshad Mobasher · 2009
Abstract. Collaborative tagging applications allow users to annotate online re-sources, resulting in a complex three dimensional network of interrelated users, resources and tags often called a folksonom A pivotal challenge of these sys-tems remains the inclusion of the varied information channels introduced by the multi-dimensional folksonomy into recommendation techniques. In this paper we propose a composite tag recommender based upon popularity and collabo-rative filtering. These recommenders were chosen based on their speed, mem-ory requirements and ability to cover complimentary channels of the folkson-omy. Alone these recommenders perform poorly; together they achieve a synergy which proves to be as effective as state of the art tag recommenders.