Sunday, October 15, 2006

Recommender Systems

Stephen Downes linked to this older (1999) page on Recommender Systems in his recent paper on Learning Networks and Connective Knowledge. It pre-dates the social-software boom, but still acts as a nice overview and notes the value of recommendations in finding people in addition to movies, books and other items:
"Although currently recommender systems are mostly used for finding things, such as books and CDs, Resnick thinks that one promising application may be recommending people. You could use recommender systems to find the right consultant or colleague - or even a potential mate."
Most people sharing a learning goal in 43 Things aren't necessarily looking for people to collaborate with (although there is functionality to form explicit groups to pursue a goal together). It could be used as a sort of recommender system for finding people ("I'm looking for people sharing my goals"), but it looks to me like it's being used more as a recommender system for things you might like to do or learn. The network that emerges around a goal does loosely connect people to each other, but that may not be as valuable as the connections between the artifacts themselves: the entries outlining what their experiences have been in pursuing the same goal, why they decided to pursue it, what they hope to accomplish, how the learning helped them, pitfalls to avoid, etc.

Perhaps it's more important as a way of finding content (advice, resources, opinions, possible applications) than as a way to find like-minded people. The primary "pivot" is the goal itself, with the people associated with each goal as secondary pivots. It is interesting to find out what other goals someone is pursuing besides the one you share with them -- that function is more exploratory than the process of figuring out if you want to pursue a specific goal.

No comments: