In recent years, contextual advertising has been widely applied to text-based web advertising. Because contextual advertising relies on text to determine the advertising context, it cannot be applied to multimedia advertisements that have no text. Therefore, this paper proposes an annotation-based contextual advertising method for matching multimedia advertisements and web pages. In the proposed method, a term vector represents a web page and an advertisement is represented by a small set of keyword tags that are either annotated by experts or extracted automatically from the text of the advertisement. This study then developed a matching mechanism for computing the similarity between a web vector and the advertisement tags. Experiments and evaluations were performed for demonstrating the performance of the proposed method. The results showed that the proposed method demonstrated a more favorable performance than traditional advertising methods.
|頁（從 - 到）||123-137|
|期刊||Journal of Electronic Commerce Research|
|出版狀態||已出版 - 2015|