http://meta.wikimedia.org/wiki/En_validation_topics
The article rating feature is going live in 1.5.
SUMMARY: Articles will be rateable on various attributes. All ratings are public and attributed, just like edits are. We'll be taking ratings from anons as well as logged-in users, since our readers vastly outnumber our editors. We're explicitly not doing anything with the data, so if 10,000 anons rate [[Image:Autofellatio.jpg]] the best article ever then it doesn't matter. For further detail, see recent extensive thread on wikipedia-l, and go to http://test.leuksman.com/ using the Monobook skin and click on the 'Validate' tab.
Now, the point of the link at the top of this message is that we haven't decided what attributes we'll be rating on. We need a good selection and discussion of them. And we need it soon - 1.5 is supposed to be rolled out early June. Presumably there will be a vote, or maybe Magnus will just pick the ones he likes. Or I will. Or something.
I particularly want to hear from academic researchers interested in Wikipedia - you folk will LOVE this data. What things would you particularly like to see reader/editor ratings of?
Also read about the feature and anticipated possible problems:
http://meta.wikimedia.org/wiki/Article_validation_feature http://meta.wikimedia.org/wiki/Article_validation_possible_problems
- d.
-----BEGIN PGP SIGNED MESSAGE----- Hash: SHA1
I would like to use this opportunity to ask developers with some time on their hands (yes, I know) to help me with the boring but necessary fine-tuning of the feature: * There's no restriction on who can edit a topic now. Should be limited to sysops or something * There's no pretty text header on the validation pages * Should the table design be changed (regarding index)?
The list goes on. You know the drill :-)
Magnus
David Gerard schrieb:
http://meta.wikimedia.org/wiki/En_validation_topicsThe article rating feature is going live in 1.5.
SUMMARY: Articles will be rateable on various attributes. All ratings are public and attributed, just like edits are. We'll be taking ratings from anons as well as logged-in users, since our readers vastly outnumber our editors. We're explicitly not doing anything with the data, so if 10,000 anons rate [[Image:Autofellatio.jpg]] the best article ever then it doesn't matter. For further detail, see recent extensive thread on wikipedia-l, and go to http://test.leuksman.com/ using the Monobook skin and click on the 'Validate' tab.
Now, the point of the link at the top of this message is that we haven't decided what attributes we'll be rating on. We need a good selection and discussion of them. And we need it soon - 1.5 is supposed to be rolled out early June. Presumably there will be a vote, or maybe Magnus will just pick the ones he likes. Or I will. Or something.
I particularly want to hear from academic researchers interested in Wikipedia - you folk will LOVE this data. What things would you particularly like to see reader/editor ratings of?
Also read about the feature and anticipated possible problems:
http://meta.wikimedia.org/wiki/Article_validation_feature http://meta.wikimedia.org/wiki/Article_validation_possible_problems
- d.
WikiEN-l mailing list WikiEN-l@Wikipedia.org http://mail.wikipedia.org/mailman/listinfo/wikien-l
On Thursday 26 May 2005 08:41, David Gerard wrote:
I particularly want to hear from academic researchers interested in Wikipedia - you folk will LOVE this data. What things would you particularly like to see reader/editor ratings of?
At first blush, it would make sense to rate articles with respect to the criteria of what makes a good article as documented on: [1] http://en.wikipedia.org/wiki/Wikipedia:What_is_a_featured_article [2] http://en.wikipedia.org/wiki/Wikipedia:Featured_articles
So it would be nice if the ratings captured whether it was well written, an appropriate reading level, an appropriate size, NPOV, the appropriate use of references, etc. However, that said, I don't think it should include anything that could be done by machine. (Lih's (2004) quality (rigor, diversity), Newberry's (2004) mass and luminosity, and Emigh's & Herring's (2005) formality don't seem to apply in this case of users' subjective ratings.) So for example, that would remove the reading level and appropriate size which could be generated automatically. Also, it should be kept relatively simple. A single subjective rating informed by [1,2] might be good enough.
On second thought, the question is very relevant to some of my experiences at the W3C. Since we advocated valid HTML, it was embarrassing that some of our pages were not valid HTML. So after time we began generating reports of the most popular invalid pages, and who owned them. This enabled us to drastically reduce the likelihood of the public encountering an invalid W3C page. So, it would be really interesting to see what are the most popular stub articles. (This to could be generated automatically from referrer, but can also be used so as to find the most popular poorly rated articles once we have that data.)
[cc'ing to wikipedia-l]
Joseph Reagle (reagle@mit.edu) [050528 02:56]:
On Thursday 26 May 2005 08:41, David Gerard wrote:
I particularly want to hear from academic researchers interested in Wikipedia - you folk will LOVE this data. What things would you particularly like to see reader/editor ratings of?
At first blush, it would make sense to rate articles with respect to the criteria of what makes a good article as documented on: [1] http://en.wikipedia.org/wiki/Wikipedia:What_is_a_featured_article [2] http://en.wikipedia.org/wiki/Wikipedia:Featured_articles
Yeah. I went through the first of those and tried to write them as rateable criteria. All improvements most welcomed.
So, it would be really interesting to see what are the most popular stub articles. (This to could be generated automatically from referrer, but can also be used so as to find the most popular poorly rated articles once we have that data.)
Indeed!
"Gather the data but don't do anything with it yet" is an idea that I think will work very nicely *because* it separates layers properly. If we create a pile of raw data, people will come up with *all sorts* of interesting things to do with it. Then maybe we can go back and tweak what we collect.
- d.