This is not news for people who've been watching closely, but I thought it deserved a "re-post" to give it some additional visibility.
In the last year, the Wikimedia movement has developed some very important content partnerships with cultural institutions such as museums and archives to bring valuable pictures, videos, and other media online. Some but not all of them are categorized here:
http://commons.wikimedia.org/wiki/Category:Commons_partnerships
What's the impact of these partnerships? How are these media used? We didn't have good answers to these questions until very recently. Thanks to the work of Bryan Tong Minh, Magnus Manske, and other engineers, we now have some first good data:
1) The GlobalUsage extension is now re-deployed on Wikimedia Commons, which makes it easy to see where any individual file is used in the Wikimedia universe;
2) The Glamorous script by Magnus Manske gives you that overview for an entire category on Commons.
For example, you can go to http://toolserver.org/~magnus/glamorous.php and select the "Images from the German Federal Archive" category. This will show you that out of the 82,457 images uploaded so far, more than 15,000 are currently used in articles. 34 languages use at least 100 images, 11 use at least 1,000. This demonstrates the powerful dynamic of global re-use that uploading media to Wikimedia Commons can result in.
We'll be able to show even more compelling data if we now add the (known) pageview data for the relevant articles. Hopefully this emerging data will contribute to a virtuous circle of new content partnerships. I'll pull together some facts for a blog update on what's happening in the space, but wanted to give a general quick update first. :-)
Erik Moeller schrieb:
What's the impact of these partnerships? How are these media used? We didn't have good answers to these questions until very recently. Thanks to the work of Bryan Tong Minh, Magnus Manske, and other engineers, we now have some first good data:
Thanks to Bryan and Magnus for teir great work on this!
We'll be able to show even more compelling data if we now add the (known) pageview data for the relevant articles. Hopefully this emerging data will contribute to a virtuous circle of new content partnerships. I'll pull together some facts for a blog update on what's happening in the space, but wanted to give a general quick update first. :-)
I think it would be even better if we could also show views for the images themselves, not only page views for the articles they are used in. It would be interesting to see how the two compare, and also, what (thumbnail) sizes of the images are viewed.
This information is not being collected at the moment, but Domas told me it would be trivial to do, if we had a place to put it. We (WMDE) are currently thinking about ways to make this data available on the toolserver, in raw for as well as aggregated figures in the database. I hope we can show off the numbers at Wikimania :)
-- daniel
On Sat, Jan 16, 2010 at 1:46 PM, Daniel Kinzler daniel@brightbyte.dewrote:
Erik Moeller schrieb:
What's the impact of these partnerships? How are these media used? We didn't have good answers to these questions until very recently. Thanks to the work of Bryan Tong Minh, Magnus Manske, and other engineers, we now have some first good data:
Thanks to Bryan and Magnus for teir great work on this!
We'll be able to show even more compelling data if we now add the (known) pageview data for the relevant articles. Hopefully this emerging data will contribute to a virtuous circle of new content partnerships. I'll pull together some facts for a blog update on what's happening in the space, but wanted to give a general quick update first. :-)
I think it would be even better if we could also show views for the images themselves, not only page views for the articles they are used in. It would be interesting to see how the two compare, and also, what (thumbnail) sizes of the images are viewed.
This information is not being collected at the moment, but Domas told me it would be trivial to do, if we had a place to put it. We (WMDE) are currently thinking about ways to make this data available on the toolserver, in raw for as well as aggregated figures in the database. I hope we can show off the numbers at Wikimania :)
-- daniel
This is probably more of a question for tech-l, but how would image-views for images placed in WP articles be tracked? We already count pageviews for articles in WP or images in commons, that's fine. Equally, we can count the clicks on an image if someone clicks the thumbnail. But, how do you measure the number of times that an image is *seen* in articles? Counting the number of thumbnail loads or the number of pageviews is not the same as proving the image was actually looked at. For example, there is a Featured Pic of the day on the front page of Wikipedia, but it is below the fold. Many many people who are loading that thumbnail are not actually seeing the image because they are not scrolling down the page.
How do we differentiate between loading a thumbnail/viewing an article and actually *viewing* an image in that article that is below the fold? Or, when we talk about image views are we only talking about those people who actually click on the image to get a larger size? (sorry if this is a question that has already been answered elsewhere).
- Liam [[witty lama]]
wittylama.com/blog Peace, love & metadata
Commons-l mailing list Commons-l@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/commons-l
Liam Wyatt schrieb:
How do we differentiate between loading a thumbnail/viewing an article and actually *viewing* an image in that article that is below the fold? Or, when we talk about image views are we only talking about those people who actually click on the image to get a larger size? (sorry if this is a question that has already been answered elsewhere).
We can't. "page views" refers to loading the page. no one can know if it's actually being looked at. Same with images. We'd just track how often the thumbnails get loaded.
Tracking the scroll state of a page, and which images are visible at what time, is possible in theory with javascript, but messy, unreliable, and generally nasty. and pointless: if the image is in plain view without scrolling, you still don't know of it's even in a visible tab or window.
-- daniel