Automatic keywords generation for contextual advertising
Pengqi Liu, Javad Azimi, Ruofei Zhang · 2014
Contextual Advertising (CA) is an important area in the industry of online advertising. Typically, CA algorithms return a set of related ads based on some keywords extracted from the content of webpages. Therefore, extracting the best set of representative keywords from a given webpage is the key to the success of CA. In this paper, we introduce a new keywords generation approach that uses some novel NLP features including POS and named-entities tagging. Unlike most of the existing keyword extraction algorithms, our proposed framework is able to generate some related keywords which do not exist in the webpage. A monetization parameter, predicted from historical search keyword performance, is also used to rank potential keywords in order to balance the RPM (Revenue Per 1000 Matches) and relevance. Experimental results over a very large real-world data set shows that the proposed approach can outperform the state-of-the-art system in both relevance and monetization metrics.