1 / 11

WebWatcher: A Learning Apprentice for the World Wide Web

WebWatcher is an information-seeking assistant that helps users find relevant links on the web. It learns from user preferences and provides recommendations based on its knowledge. With WebWatcher, users remain in control of the search process while receiving valuable suggestions.

laurenced
Download Presentation

WebWatcher: A Learning Apprentice for the World Wide Web

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. WebWatcher: A Learning Apprentice for the World Wide Web Robert Armstrong, Dayne Freitag, Thorsten Joachims and Tom Mitchell 발표자: 자연언어처리연구실 김정집

  2. M o s a i c WWW WebWatcher User Overview • WebWatcher • information seeking assistant for the WWW Advice Prefetch Tracking user’s search

  3. Typical scenario • Typical scenario 1. User sets up WebWatcher for his/her goal. 2. WebWatcher now “looking over his shoulder” and recommend related link by its knowledge. 3. User decides to select this advice or not. 4. Until the user dismiss WebWatcher continue step 2~4. • User remains firmly in control.

  4. Typical scenario

  5. Typical scenario

  6. Typical scenario

  7. Learning for guiding search(1/2) • What Should be Learned? • Simple learning function • UserChoice?:Page Goal Link->[0,1] • the value is the probability that an arbitrary user will select Link given the current Page and Goal • How Should Pages, Links and Goals be Represented? • 530 boolean features

  8. Learning for guiding search(2/2) • What Learning Method Should be Used? • Winnow • Wordstat • TF/IDF with cosine similarity measure • Random • to provide baseline measure

  9. Results(1/2) • 30 sessions using WebWatcher to search for technical papers • How Accurately Can UserChoice? be Learned?

  10. Results(2/2) • Can Accuracy be Improved by Sacrificing Coverage?

  11. Conclusions • Software assistance is needed. • The accuracy of the agent’s advice can be increased by allowing it to give advice only when it has high confidence. • We are optimistic that a learning apprentice for the WWW is feasible.

More Related