Improving contextual advertising by adopting collaborative filtering

Abstract

Contextual advertising can be viewed as an information filtering task aimed at selecting suitable ads to be suggested to the final “user”, that is, the Web page in hand. Starting from this insight, in this article we propose a novel system, which adopts a collaborative filtering approach to perform contextual advertising. In particular, given a Web page, the system relies on collaborative filtering to classify the page content and to suggest suitable ads accordingly. Useful information is extracted from “inlinks”, that is, similar pages that link to the Web page in hand. In so doing, collaborative filtering is used in a content-based setting, giving rise to a hybrid contextual advertising system. After being implemented, the system has been experimented with about 15000 Web pages extracted from the Open Directory Project. Comparative experiments with a content-based system have been performed. The corresponding results highlight that the proposed system performs better. A suitable case study is also provided to enable the reader to better understand how the system works and its effectiveness.