An Enhanced Model for Behavioral Targeting in Online Advertising

With the wide popularity of the internet, online advertising has become a major method for marketing. Currently all the major online advertisement providers (including Google Adsense and Facebook) implement Behavioral Targeting (BT) as their online advertisement strategy. One of the major drawbacks of BT is that the advertisement delivery considers only the history of a user's browsing behavior; the relevance of the advertisement to the web page is totally ignored. There have been attempts to bridge the gap by combining contextual relevance with BT. But contextual relevance being a static measure has the problem that it does not adapt to user's click behavior. We propose a model that combines both user's interests and the interests of all users who have visited the page. This model is tested using data published by Chinese search engine Sogou. The model performs better than pure BT.