On detection of contextual advertisements

Web advertising has become a major industry and a large part of this market consists of contextual ads. Although it has made a great impact on earnings of many publishers' websites, these advertisements tend to disturb the internet surfing of normal users and to consume a lot of valuable bandwidth. Moreover, they always bring extra burden in indexing to commercial search engines as they mix up with the main content of the hosting web pages. Therefore, it is necessary to automatically detect those contextual ads on the web. In this paper, a classification based approach is proposed for contextual ads detection. Those features include text, link, layout and style in hosting web pages. Furthermore, neural network is used to identify the parameters that contribute the most in detecting contextual ads from non-contextual ads. Promising experimental results are obtained on ATOM textual snippets collected from 219 web sites.