[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
GroupLens, WWW-Rating
- To: Fitug-Debatten <debate@fitug.de>
- Subject: GroupLens, WWW-Rating
- From: Thomas Roessler <roessler@guug.de>
- Date: Tue, 3 Mar 1998 23:04:53 +0100
- Comment: This message comes from the debate mailing list.
- Mail-Followup-To: Fitug-Debatten <debate@fitug.de>
- Sender: owner-debate@fitug.de
Unten angehängt ist
http://www.cs.umn.edu/Research/GroupLens/grouplens.html.
Wenn man den Gedanken weiterspinnt auf ein positives _wie_
_negatives_ Rating von WWW-Seiten, könnte man zu recht
interessanten Ergebnissen kommen: Die Kriterien des
Ratings sind letztlich in der Korrelation mit anderen
Ratenden verborgen und dadurch sehr viel differenzierter
als etwa ein rsaci-Rating. Kultureller Hintergrund wird
prinzipbedingt in das Rating mit einbezogen, ebenso die
persönlichen Interessen.
Problematisch ist natürlich die benötigte Rechenpower -
und auch der liebe Datenschutz...
Schaut's Euch einfach mal an.
tlr
------------------------------
The GroupLens Recommendation System
_______________________________________________________________________________
[1]How to Rate Articles
[2]Register for the GroupLens Service
[3]Download GroupLens Software
[4]GroupLens Home The GroupLens System.
The GroupLens system is an article recommendation system for
electronic forums, specifically Usenet news. The purpose of GroupLens
is to increase the value of time spent reading electronic forums.
Internet newsgroups can carry hundreds of new postings every day. Many
of these articles are off the newsgroup topic, and many more are not
personally interesting to you. It is no longer feasible to read every
article posted to a newsgroup in order to find interesting content.
The GroupLens system makes reading Internet news productive again by
highlighting articles of likely interest and warning against articles
that will not be interesting. Using the GroupLens system, you won't
miss the good articles and you won't waste your time on the bad ones.
When you first see an article in a newsgroup, chances are that it has
already been read by lots of other people. Each of these people formed
an opinion of how interesting the content of that article is. Imagine
you could talk to each of these people and could ask them for their
opinions of the quality of the article. Based on these
recommendations, you would determine if it was worth your time to read
the article. Of course you can't talk to all those people, but the
GroupLens system can, and all you have to do is tell GroupLens your
opinion on articles you read. Users of the GroupLens service are asked
to rate each article on a scale of 1 to 5, based on how interesting
they found the article. The GroupLens system takes everybody's ratings
and matches people who agree frequently. When you first see a new
article, the recommendation system uses the ratings of people who
agree with you to generate a prediction of your interest in the
content of the article.
Currently, the GroupLens system provides two free services to the
Usenet community.
* Personalized Predictions. As you rate articles, your ratings are
compared with other users' ratings on the same articles. You are
matched with people who repeatedly agree with you or repeatedly
disagree with you. Ratings from these users are then use to
generate a prediction reflecting your personalized interest.
Personalized predictions are provided for a selected subset of
[5]groups.
* Average Predictions. Due to computational limitations, we can't
provide personalized predictions to all newsgroups. We recognize
that you may be excited about using the GroupLens system, but not
read any of the supported groups. So for all other groups, we
provide predictions that are not personalized but are the average
rating. These are especially valuable in groups which contain many
articles that everyone agrees are bad or good, such as groups with
frequent flame wars or frequent spam attacks.
Motivations
The GroupLens service is part of the ongoing GroupLens research
project at the University of Minnesota Department of Computer Science.
It is being provided both as a public service and as a means to gather
research data on personalized recommendation systems. We have no
intention of charging money at any point in time. We are committed to
providing it for 3 years, so you can depend on it being around for a
while.
Further Information
[6]How to rate articles? (What should get a 5?)
[7]How Does GroupLens Work?
[8]Register for the GroupLens service
[9]Download GroupLens-enhanced software
[10]Original GroupLens announcement
This page was last updated on: June 11th, 1997
Questions or comments? [11]Please send us mail.
References
1. http://www.cs.umn.edu/Research/GroupLens/rating.html
2. http://www.cs.umn.edu/Research/GroupLens/registration.html
3. http://www.cs.umn.edu/Research/GroupLens/software.html
4. http://www.cs.umn.edu/Research/GroupLens/index.html
5. http://www.cs.umn.edu/Research/GroupLens/supportedGroups.html
6. http://www.cs.umn.edu/Research/GroupLens/rating.html
7. http://www.cs.umn.edu/Research/GroupLens/devGLIntro.html
8. http://www.cs.umn.edu/Research/GroupLens/registration.html
9. http://www.cs.umn.edu/Research/GroupLens/software.html
10. http://www.cs.umn.edu/Research/GroupLens/announcement.html
11. mailto:grouplens-info@cs.umn.edu
--
Thomas Roessler · 74a353cc0b19 · dg1ktr · http://home.pages.de/~roessler/
2048/CE6AC6C1 · 4E 04 F0 BC 72 FF 14 23 44 85 D1 A1 3B B0 73 C1