Mine The Web’s Socially-Tagged Links: Google Social Graph API Launched

Did you know there’s a way to tag links on pages to indicate social connections? I’d heard about this vaguely, but you can bet there’s going to be much more public awareness and potential use, thanks to Google launching its new Google Social Graph API. Now available, the API allows developers to discover socially-labeled links on pages and generate connections between them.

"We want to make the connections searchable. We think there are a few things people will do with that. We think they’ll build some [social linking] exploration tools," said David Glazer, a director of engineering at Google who oversees the company’s OpenSocial effort.

The API lets developers locate links that make use of the XFN and FOAF labeling systems. XFN is pretty easy to understand. If I wanted to link to my personal blog and indicate it was mine, I could do it this way:

Daggle

<a href="http://daggle.com" rel="me">

See the code underneath the link? That’s the code used to make the link. The rel="me" part says that I’m pointing at something that is me, i.e., that the author of this page says belongs to me. There is a list of 15 types of relationships that can be defined, such as someone who is a co-worker or someone you have a crush on.

Very few people use these labeling mechanisms, Glazer said. However, they exist out there in a variety of tools that add them without you knowing.

For example, here’s my Twitter profile:

dannysullivan

On it, you’ll see that my profile box lists this for my web address:

Look at the code in that link, and you’ll see Twitter automatically labeled the link as "me" without me even realizing that. Other tools like LiveJournal, WordPress, and Vox do similar labeling.

Developers can harvest this data and play it. Glazer stressed that Google has no idea what will be done with it. The company is just putting the ability to mine its index out there and seeing what comes of it.

Even if you’re not a developer, you can have some fun playing with the API. There some forms you can use to explore page connectivity here. Note that they aren’t working properly for me right now, but that will likely be corrected soon.

In the example below, I entered the home page URLs for my work site (searchengineland.com), my personal blog (daggle.com), and my Twitter profile:

It came back to tell me based on link data that Twitter was explicitly saying it was connected to Daggle (so 2 out of my 3 sites were connected there). But it couldn’t find a connection between my sites and Twitter (which is the case). I could easily correct that and build up my linkage if I wanted to. I haven’t because, at the moment, there’s been no compelling reason to do so.

Note that it also found other sites claiming a connection to me, such as my FriendFeed, LinkedIn, My Blog Log, and Technorati profiles. These are all again places (I assume) where because I’ve listed URLs, the "me" attribute has been added. Another tool can show those who link to me as a contact.

My chief concern was what happens if someone wants to "fake" being me. Glazer said this is difficult because of the number of links I can control will tend to outweigh anyone else. Being dubious of meta data in general, I still want to go hmm, we’ll see. But since we don’t even know what apps will make use of this data, it’s a bit early to worry.

One chief advantage that Glazer said that may emerge is that if people do more tagging, it may be easier for them to go to any social platform and have their profiles be autofilled. Indeed, even if they don’t tag, this could happen. For example, you might sign up for a Twitter account, enter your blog URL, and Twitter could use the API to scan for anything connected to your blog (such as LinkedIn, etc.) and prepopulate fields. It’s something that Vanessa Fox wrote a long post about wishing could happen, and perhaps it really will happen.

For more, see discussion on Techmeme here.