Jon Udell: Collaborative filtering with del.icio.us (and other ideas)

Jon Udell: Collaborative filtering with del.icio.us

That coolness factor of browsing and finding other people’s del.icio.us pages because they seem to share a common interest, could be, in part, an automatic process. Here John introduces a way it could be done, very very cool!

I have been filling in key-words of my interests like, for-ever, in profile pages all over the internet. And I dont think that info has ever spawned an online relationship with anyone else..

With the approach John shows here, we can go beyond these superficial keywords, and drill down deeply into a persons interest space. The only downside to this is that where are still being rather virtical here, as in we must bear in mind that we are only looking at a persons bookmarking behaviour, which says nothing of his information-consumption behavior like what he browses, what feeds he aggregates.

Though of course the latter can be analyzed if one uses a service such as bloglines.

If you where to take my del.icio.us page, and compare it to my bloglines page, you get a far more nuanced view of what I do online, and where my interests lie. The next step thus, is to aggregate that information together, and there are many more sources that one could reference in this manner. My blog for instance. Or my personal website. And then there is other kind of meta-info on me, like my online reputation on sites like ebay, slashdot, tweakers.net, or how about online CV’s or forum activity, or comment activity?

What if one would create a crawler application that, using all of these sources, could compine a complete ‘view’ of my digital online self?

The problem though, is that ‘social’ internet tools, that effectively visualize this stuff, are not all that common yet. Bloglines, Del.icio.us, and Flickr, and a few others, are still few and far between.

And then there is privacy. Now personally, I have no problem sharing all this metadata about myself. I would not even mind my browsing habbits being monitored to service the ‘social’ metaspace like I do with the above-mentioned services, provided I had complete and transparent control over when what was monitored.

Wouldnt it be kinda creepy, once we indeed had a centralized match-and-compare system for all of this data, if you where to run into another person online that mirrored you and your interestes in every way? And online doppelganger, so to speak


Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.