The presentation of personalized information is a research area that has gained much impetus from the growing number of Web information resources. The system discussed in this paper tackles the issue of personalizing news services.
From a number of third-party news services, the system selects news articles that may be of interest to a user, selects the level of detail at which the article should be displayed, and also selects appropriate advertisements for each user and page. The decisions are made based on inferences from a user model. This system uses a probabilistic rule-based architecture for representing a user’s interests and personal details. Personalization is through both the content of articles, and user characteristics such as the user’s, prior knowledge of a particular area.
Although the system makes use of a wide variety of information, it is the user who is being modeled, rather than the combination of user and information resources. The paper presents a complete, mature system rather than the ongoing development of a new theory. As such, the only disappointments are the lack of a rigorous user evaluation and detailed examples of the system in use.