Rich blog content at the click of a button – Zemanta has gone live!

Source: Wikipedia

This is great news! In February, I wrote for the first time about the Zemanta browser plug-in prototype which was supposed to allow you to enhance your blog’s content by automatically suggesting links (e.g. to Wikipedia or news pages) and pictures (e.g. on Flickr), based on semantic analysis of your text. Today Andraz from Zemanta notified me that they went live – and the working version is even cooler than the demo they had on their website in February: In February, you had to enter text and hit a button to ‘zemify’ the text – but the current Zemanta comes as a WordPress plug-in. Every 300 characters as you type the plug-in suggests further links and tags which you can apply all at once or by clicking on each one you want – and that is of course MUCH MUCH more convenient than going to a website, copying the URL, highlighting the word and hitting the link button in WP.

The next cool feature is that a side bar shows related content (articles and pictures) on the web – which you can simply add to your blog post by, again, simply clicking on them. Extremely cool! And yes: This very blog entry has been enhanced with Zemanta!

But of course there are a number of glitches in this early version… (and how wouldn’t there be any, as they need feedback to improve):

• The amount of code that is added to your blog entry is a bit intimidating. Adding links runs smoothly, as those links are added to text you’ve already written – but it would be cool to be given the opportunity to decide whether one wants their ‘rich’ links (with title, relation and class attributes) or just plain ones.
• A little footer, explaining that the text was enhanced with Zemanta, is automatically added. Might sound fair enough – but it would be nicer if that were a voluntary option and didn’t happen automatically. Of course you can still remove that footer – but having to opt out is not the same as voluntarily opting in.
• Adding a picture also automatically adds more code (span tags with e.g. margin attributes) than seems necessary – I’d rather have a plain image link (where I can easily define the size of the image myself by specifying width and alignment) than those six (!) lines of code that are pasted into my editor at the moment. It just takes far too long to scan this huge amount of information to see how it can be altered.
• The quality of link and tag suggestions can still be improved (of course); ‘URL’ or ‘the web’ and actually not even ‘Flickr’ (because everyone knows Flickr, right?) are not a tremendous enhancement in the link department. Maybe they’ll be able to track which suggestions are used and which aren’t, and are thus able to identify the empty ones (in the same way that Google disregards ‘a’ or ‘the’ in searches). What’s going to be more difficult is to automatically identify which PHRASES should be linked – as most people (I think? I do) tend to link phrases, and not just single words. Most links in this post, and all pictures, were automatically added by Zemanta.
• Final suggestion: The image gallery is nearly to small to assess whether the suggested images are suitable or not – but I wouldn’t know how to make them much smaller really. What’s great though is that it automatically shows which pictures you have already used and which you haven’t – so that you can automatically remove the used ones by hitting ‘minus’ in the gallery preview. At the moment, the gallery suggestions only vaguely correspond to my content – it’d probably easier if I wrote about kitten and horseys though.

That much about my feedback. I’ll keep testing Zemanta for while, and am keen to see how it is going to improve over the next weeks!