I recently spoke with Steve Ganz, LinkedIn's Principal Web Developer, about the role of semantic technologies in LinkedIn's strategy.
SR: Steve, I've seen some press releases recently about how LinkedIn is using semantic technologies as part of its strategy. When did this start?
Ganz: LinkedIn has been using some aspects of semantic technology since 2006. Mainly we use a variety of microformats in our web presence. In fact, LinkedIn is the largest publisher of the hresume format.
SR: Do you distinguish between front end and back end uses of semantics?
Ganz: Most of what we use is what I categorize as "lower-case" semantic web, which is different from the "upper-case" Semantic Web technologies. Our uses are almost entirely based on the use of microformats in the front end. We use them to give the user access to their own data in ways beyond simply viewing it on the website. I like to think of microformats as the "format for the people." They are easy to implement by applying meaningful class names to well structured, semantic HTML which is aready marking up existing published human readable data.
SR: When I was on my LinkedIn account the other day, I noticed a section titled "People you may know." As I was thinking about semantic technologies, I made a guess that there was some kind of semantic analysis going on in the background to enable this. Is that the case?
Ganz: We don't actually use any semantic structures in our data that I'm aware of. The "People you may know" section is created by using sophisticated algorithims on traditional data.
Our concentration is on making the users data easily available and useful. To that end, we use a variety of microformats including hCard, hResume, hEvent, and hReview. We are also implementing the XFN microformat (XHTML friends network), which is one of the cornerstones of the Google Social Graph API.
This works by applying the “rel” attribute with a value of “me” to links to your other websites, then the Google spiders go out and discover all of your related sites. Our public profile is a good starting point for that as we move toward exploring OpenID and OAuth for digital identities.
SR: How do you see the adoption of microformats in general?
Ganz: Microformats are gaining in adoption from the publishing standpoint. It's practically ubiquitous compared to some of the upper-case semantic technologies. Yahoo! is a huge publisher of microformats, and Google now publishes geo locations in microformats. It's definitely gaining popularity.
I think this is because using microformats doesn't require building any back end infrastructure. Rather, it's easy to implement for the front end web developer, and they will naturally gravitate to it.
SR: Do you see what you refer to as the upper-case and lower-case semantic communities coming together?
Ganz: There are definitely benefits to having the two communities work more closely. It hasn’t always been that way, but I sense a bit more of a cooperative spirit now.
SR: What's next as far as LinkedIn's use of semantics?
Ganz: We will hopefully be experimenting with RDF and FoaF in the coming months. As the technology becomes fully realized, I'm sure that the Semantic Web will be every bit as profound a development as the WWW was.