Audrey IBM Research

Audrey is a social news service that is automatically tailored to each user's interests. Audrey makes it easy to share and discuss interesting news stories with friends and colleagues. RSS were continuously polled, and natural language processing was used to extract disambiguated named entities and their relationships from the text. These graphical models were generalized to created "interest profiles" for each user that were used by an automatic recommender system. The site featured a Twitter-like follower model, an activity stream, and other social features.


Steven I. Ross, Daniel M. Gruen, Todd Soule, and John F. Patterson


Java, Dojo, OpenCalais, and PostgreSQL