On 9/30/05, Amir Michail <[EMAIL PROTECTED]> wrote: > On 10/1/05, Scott Villarosa <[EMAIL PROTECTED]> wrote: > > I think this is it. > > > > Well, if that's the case, then how far do you think we can go in > analyzing del.icio.us users? > > Obviously, certain things would be unacceptable such as automatically > inferring an estimate of their IQ based on their bookmarks! > > But what about inferring their personality? Is that going too far? > What about their profession? Their true identity?
I was only just starting to think of this one the other day: If you have a way to watch users using del in action, you could observe their categorizing performance under conditions of varying distraction. (operationalize the performance by looking at number of tags they use when in a hurry, at work, etc (semantic distractions vs non-semantic) as well as the seeming "fit" of the tags to what they are bookmarking though that is much more problematic) That has probably been done before in many, many experiments but the cool thing is that del is not in an experimental setting and there is such a huge population of users, and they are voluntarily doing tasks that in another setting an experimentor might have had to trick them into doing. Of course the huge problem is that no one declares intent when using the service--so this is like observing animals in the wild vs. in a lab. :) del is like a huge memory experiment, as well as a huge meta-cognition experiment. In school I remember participating in meta-cognition experiments where they had pre and post surveys about expected performance in a class, how many hours one studied for a test, how one exected to do, etc. It would be hard to figure out the "successful use" of a stored link, but the common intent is to save something to find it later (duh?) and we get to watch how people think they'll be able to find their information later. they'll make up a list of key words, and depending on how they use del use some sort of combination of recall, recognition, etc. You could take a population of del users and have an almost guaranteed set of subjects who have decided to remember something. You could study recall, recognition. look at the affect of context, infer the level of processing they did based on the tag set (except this is clouded now by suggested tags, but otoh you could use this data to improve suggested tags?) and then study performance based on level of processing. just thinking out loud. -- sheila _______________________________________________ discuss mailing list [email protected] http://lists.del.icio.us/cgi-bin/mailman/listinfo/discuss
