| A recent surge of subscriptions to online news services exemplifies the fact that people and organizations constantly need up-to-date information to stay competitive and make better informed decisions. However, these news services often have poor service quality due to lack of personalization and intelligence. This causes these services to constantly overload consumers with irrelevant information. In this dissertation, a new two-stage model for personalized information delivery is developed. In particular, the two-stage model first helps users better represent their information needs by formulating individually-tailored Persistent Queries (PQs), then it uses genetic programming to discover ranking functions that effectively combine the terms contained in the PQs. The two-stage model is empirically validated using a widely used very large text corpus. |
