The Personalized News approach to news aggregation relies on pattern recognition in order to deliver user/readers news they will find relevant (based on their previous indications of what they find relevant). The idea is that everyone has their own pattern of reading. To recognize your pattern, personalized news (sometimes called personalized content) services omit stopwords and extract keywords from the news you read - then use a statistical analysis to predict what kind of news you will like or dislike in future.
- Personalized News: A Market Overview from the Read/Write Web
- Personalized Content vs. Power of Masses: Which Will Prevail? from Digital Alchemy