An intelligent agent for Web advertisements

The rapid growth of Internet users attracts advertisers to post their advertisements on the Internet. The probabilistic selection algorithm was not satisfactory; while other advertising agents are unable to guarantee the quality due to insufficient and unstable user information. We describe a new advertising agent based on user information. The users' interests are discovered by the order pattern mining algorithm first, then applied to the Gaussian curve transformation to represent their profiles. For the advertisements, we use the keywords from different categories to construct the advertisement profiles as Gaussian curves also. This allows us to select advertisements based on the intersections of the different profiles according to the users' preferences in an effective and efficient mechanism. A prototype of the Intelligent Advertising Agent has been developed with Java and Oracle. From our evaluations, we observed that about 80% of the test cases are successful in making predictions which generated the most favorable category that the users are interested.