next up previous
Next: What Should be Up: WebWatcher: A Learning Apprentice Previous: WebWatcher

Learning

The success of WebWatcher depends crucially on the quality of its knowledge for guiding search. Because of the difficulty of hand-crafting this knowledge, and because we wish for many different copies of WebWatcher to become knowledgeable about many different regions of the Web, we are exploring methods for automatically learning this search control knowledge from experience.





Thorsten Joachims
Thu Feb 9 16:27:34 EST 1995