Welcome to Particletree!

After a 48 hour marathon of honest web development (minus a IMAX Batman Begins break and a chicken fingers meal), Particletree is proud to submit our Y!Q Challenge implementation. We call our baby “Lazy Linking” and the system is what’s generating all the purple Y!Q search links on this site.

Lazy Linking is for the sloth blogger in you that’s tired of having to find interesting links to pepper your blog posts to make them more interactive. Essentially, Lazy Linking finds the keywords in your post and turns them into Y!Q search links “automagically”.

The clever part of our implementation is that it’s integrated with our content manager, Textpattern (although it could be easily adapted for Movabletype or even WordPress). It creates minimal impact on the accessibility of our pages thanks to some clever unobtrusive JavaScript and it won’t overwrite your current links thanks to some smart php/regex action under the hood.

Because we’re a huge fan of Textile and Markdown, we made it a requirement that using Lazy Linking on our site shouldn’t affect the text readability we get from using such humane text filters. So our functions are called on AFTER Markdown does it’s magic in Textpattern, but before going into the mySQL database.

Basically, we wrote a php function to take the html formated post (after Markdown is done with it) and send it up to Yahoo Term Extraction to give us their interpretation of the post’s keywords. We then take those keywords and compare them with any keywords we might have already added in Textpattern. We join the two arrays together, remove any duplicates and update the keyword column in the mySQL database. Those keywords are then used by a smarter function to determine what words or phrases in the post should be Y!Q-ized.

What’s nice about this is that if you don’t feel like coming up with keywords/tags for your post, you don’t have to — just hit save and Lazy Linking will have Yahoo do it for you AND you’ll see them in your keyword field for your post. If you do have keywords, it’ll just enhance the Y!Q Lazy Linking experience.

We want to also emphasize that the term extraction (keyword determination) isn’t done everytime a user looks at a page or entry. It only does it when you save your post in Textpattern. This method keeps your users from using the maximum allowable 5000 calls to yahoo’s server very quickly.

If you take a quick view of the source on this page you’ll notice the only markup designating a lazy link is a ‘span’ tag designated as a ‘yqlink’ class. The other ten lines of form and input code that Yahoo requires to make each Y!Q link work is being put in thanks to some clever unobstrusive JavaScript. This makes it very easy for us if we ever wanted to intentionally make a word or phrase lazy linked—just add those span tags and the javascript will handle it for us. The nice thing about this method is that it makes creating a contextual search VERY easy and it keeps our markup looking clean. Because they’re just span tags instead of links, they shoudn’t ruin search engine crawling and ranking, because those little bots won’t see a link, just words with spans.

Other features that are nice about our implementation:

• Doesn’t replace links we’ve created.

• Doesn’t lazy link more than three words per paragraph so it doesn’t look excessive.

• Lazy links don’t show up in your rss feed (this prevents the forms needed for Y!Q search to work frommessing them up)

• Ensures that every Yahoo keyword is lazy linked at least once

• Implements some selective randomness to ensure links are organically distributed throughout a post.

• You can easily tell it what HTML elements should not be Lazy Linked. (For example we don’t want it to do headlines, cites or code tags).

• We added some pretty-making JavaScript to make the link color results cycle through a set of colors to make them more interesting to look at and match our site design.