As the title suggests, this paper describes an algorithm for rating-based collaborative filtering. This system calculates what readers like separately from what readers dislike. Such systems play an inmportant role in content filtering, and content filtering is a more robust and efficient mechanism for content selection than is a metadata based keyword search. It is the existence of work like this that gives me confidence in the use of aggregator technologies rather than search-based technologies for learning object syndication. Via Seb Paquet.
Today: 2 Total: 107 [Direct link] [Share]
