Topic detection and interest tracking in a dynamic online news source

Digital libraries in the news domain may contain frequently updated data. Providing personalized access to such dynamic resources is an important goal. We investigate the area of filtering online dynamic news sources based on personal profiles. We experimented with an intelligent news-sifting system that tracks topic development in a dynamic online news source. Vocabulary discovery and clustering are used to expose current news topics. User interest profiles, generated from explicit and implicit feedback are used to customize the news retrieval system's interface.