On smart and accurate contextual advertising

Publication Type:
Conference Proceeding
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), 2012, 7240 LNCS pp. 104 - ?
Issue Date:
Filename Description Size
Thumbnail2013002767OK.pdf51.31 kB
Adobe PDF
Full metadata record
© Springer-Verlag Berlin Heidelberg 2012. Web advertising, a form of advertising, which uses the World Wide Web to attract customers, has become one of the most important marketing channels. As one prevalent type of Web advertising, contextual advertising refers to the placement of the most relevant commercial ads into the content of a Web page, so as to increase the number of ad-clicks. However, some problems such as homonymy and polysemy, low intersection of keywords, and context mismatch, can lead to the selection of irrelevant ads for a generic page, making that the traditional keyword matching techniques generally present a poor accuracy. Furthermore, existing contextual advertising techniques only take into consideration how to select as relevant ads for a generic page as possible, without considering the positional effect of the ad placement in the page. In this paper, we propose a new contextual advertising framework to tackle problems, which (1) uses Wikipedia concept and category information to enrich the semantic representation of a page (or a textual ad) and (2) takes the placement position of embedded advertise into account. To accomplish these steps, we first map each page (or ad) into three feature vectors: a keyword vector, a concept vector and a category vector. Second, we determine the relevant ads for a given page based on a similarity measure which combines the above three feature vectors. In dealing with position-wise contextual advertising, the relevant ads are selected based on not only global context relevance but also local context relevance, so that the embedded ads yield contextual relevance to both the whole targeted page and the insertion positions where the ads are placed. We experimentally validate our approach by using a real ads set, a real pages set, and a set of more than 260,000 concepts and 12,000 categories from Wikipedia. The experimental results show that our approach performs better than the simple keyword matching and can improve the precision of ads-selection effectively.
Please use this identifier to cite or link to this item: