suggest pictures, links, articles, and tags related to your blog postings. Using proprietary natural language processing and semantic algorithms, Zemanta compares the words in a blog post to their pre-indexed database of other content in order to suggest related items which will display next to your blog post.
The articles Zemanta suggests come from 300 or so “top media sources” as well as the other blogs of Zemanta users. The images suggested come from Wikimedia Commons, flickr, and stock photo providers like Shutterstock and Fotolia.
In their FAQ they say All content that we are suggesting is copyright cleared – either clearly licenced as Creative Commons, or approved by stock providers.
I am happy to see such a service provided. I did think there was a space for it some time ago. If they do it right, then hey, win-win.
As I don’t have a Blogger, Wordpress or Typepad blog, this is a request: if you have one of these blogs, please try it out and let me know what it’s like.
- Where are the images hosted? Are they hotlinking from Commons, hosted on their own servers, or they do some voodoo and a local copy gets uploaded as part of your blog? Are there options about file sizes?
- Can you get SVGs?
- When you use it can you easily see which source the image results are from? Is it possible to restrict them to only use one image source?
- Does it seem that they are indexing all of Wikimedia Commons, or only Creative Commons-licensed content? (try to find and insert a GFDL image, or public domain)
- What is the attribution like? Are they able to extract the author information? Do they provide an automatic link back to the Wikimedia Commons page? Do they provide the license name and link?
I installed it to check. In order, then:
They are hotlinking from commons.
You can get SVGs.
You can’t see where it comes from without inserting it and inspecting the code/attribution.
They are indexing all of commons, I think, since I got a PD image (helpfully labelled as PD by the extension, which is nice).
The attribution is bad. It’s just “Source: Shutterstock” or “Source: Wikipedia”, not even “Source: Wikimedia Commons”. They don’t list the author or the license, though they do link back to the image page.
It’s kind of neat, but the hotlinking and poor compliance with licenses are rather serious drawbacks, IMO.
— Tracy Poff · 28. March 2008, 13:30
Thankyou Tracy! I hope we can help them improve the attribution before it becomes too widely used…
— pfctdayelise · 28. March 2008, 14:17
I have already seen this “Source:Wikipedia” thing. One of our news server is using this type of attribution, two years back I remember. But it is unlikely to be connected to this extension anyhow.
— Zirland · 28. March 2008, 16:35
Hy from Zemanta!
Glad to read review on your site!
Well, I think it is best to get answers from the first hand, so here they are: – currently we use the following sources for images: Wikipedia, Flickr, Fotolia, Shutterstock
While we have some of this databases at hand, some others do not give out all the information needed for attribution.
For Wikipedia we do our best to inform the author about what is the license of the image and link to the image description in order for him to be able to judge all the necessary details. It is impossible for computer to judge weather the picture can be used or not (for example when using things according to fair use)
Also wikipedia does not provide very good descriptions that would enable correct attributions (trust me, there is so much noise there right now). We are hearing from wikipedia guys they are really eager to clean that stuff up, and as they will do it we will also use and present that information.
- if you click on the image, license becomes visible under the gallery. We plan to have this ‘on hover’ soon, so it will be easier to inspect it.
- linking images directly. It is really hard problem since most users don’t really have uploading setup correctly (most local wordpress instalations, quota problems, etc.). This is the current solution and if this is going to be requested feature we will try to do ‘black magick’ and upload stuff automatically to the blog. But this means other security implications that need to be resolved.
We are really trying to be the good guys here. And we are deeply appreciating your worries about attribution.
I am sure you have many more ideas and I would welcome you to share them with us via email or via http://getsatisfaction.com/zemanta where we can discuss it with many users at once.
If you have an idea how to correctly capture wikipedia images attribution (something that would assure at least 50% correct coverage from 2.8M images), please help us! ;)
— andraz · 28. March 2008, 21:59
I’m a big fan of Z… it makes the blog authoring process faster and easier. I’ve started participating at their GetSatisfaction support area and they seem very responsive and friendly… highly recommended
— Lar · 15. April 2008, 10:21
Elsewhere on the web...
Commenting is closed for this article.