March 16, 2003
Lynching Information
Saw Clifford Lynch speak over at UC Berkeley on Friday, courtesy of Ethan Eismann. Recently reformed OG blogger Peter Merholz also showed up.
Lynch took way to long to get to the interesting bit of his talk, research into collaborative filtering and personalization. Talked ended just as he started getting to the juicy issues. Did get a chance to talk to him about the issues of information segregation though.
The issue was one he new about from discussions of online newspapers, but still remains unanswered. He did note that when people configure online news sources they general will select very focused interests and then balance that out with some sort of filtered general list like the top Reuters feeds. So at least part of the answer lies in picking good filters that know how the mix up the info flow the way a good newspaper does.
Another interesting (but discouraging) part of the talk was his reference to problems in getting access to large enough groups of people to test out collaborative filtering ideas. More anecdotal evidence of the inverse Metcalfe's Law?
Posted by William Blaze at March 16, 2003 05:21 PM | TrackBackInformation segragation is the same as information filtering, no?
Something I've been interested in is a 'serendipity slider.' Personalized news feeds filter out articles that are otherwise provided, when you browse for news. A serendipity slider allows the user to receive articles about increasingly less relevant topics, or from diverse sources. For example, if my personalized news feeds only show me articles about "war" and "blogs", setting my serendipity slider to high would occasionally give me articles about "peace," &tc. Perhaps the serendipity slider is configured to give me articles about topics I've assigned secondary importance. Perhaps it assigns me topics determined by a process of latent semantic analysis. Or perhaps the slider collects articles about popular topics, as indicated in Google News.
Posted by: Ethan on March 18, 2003 01:58 PMSerendipity slider, I like it, sounds like a great interface/algorithm
"Information segregation is the same as information filtering, no?"
No they are quite different actually. There is a big difference in scale. Filtering can lead to segregation if applied rigidly over time, but it also has huge beneficial uses in individual situations and when used properly. Information Segregation is a potential long term negative side effect of badly applied filtering.
Posted by: William Blaze on March 18, 2003 05:52 PM