Multi-agent System for Web Advertising

The main aim of a personalized advertising system is to provide advertisements, which are most suitable for the given anonymous user navigating the web site. To achieve this goal, many sources of data are processed in one coherent vector space: the advertisers’ and publisher’s web site content, sessions of former users from the past, the history of clicks on banners and the current user behavior as well as the advertising policy promoting certain campaigns. The multi-agent system running mostly on the publisher site is introduced to organize personalized advertising. Each cooperating agent is responsible for a separate, specific task: web content and usage mining, click-through data exploration, user monitoring, advertisement recommendation and management.