FOA Home | UP: Other learning applications and issues


Adaptive Lenses

A discussion of the routing or filtering task should make it clear that learning technology can be placed at many levels within the FOA system. A personal classification tool can be very useful to a single individual organizing his or her own Email. These categories need not make any sense or be consistent with anyone else. But when multiple users all browse through shared corpora, it becomes possible for one person's browsing experience to benefit another's. Of course, inter-user consistency in RelFbk will help to determine just how statistically correlated these training signals are. But if users are clustered as part of socially cohesive groups (for example, a research lab full of students and faculty pursuing the same research questions) and when they are searching documents of shared interests (for example, reprints they have all collected on topics of mutual interest, as part of a journal club perhaps) it is not unreasonable to believe that their assessments will be very similar indeed. Figure (figure) shows a single individual with a classification system on their own machine. But their searches often go through a second classifier or ``adaptive lens" that they share with other members of a computer science group. The figure also shows two different social groups of users, for example, computer scientists vs. cognitive scientists, each learning different connotations for the phrase {\tt NEURAL NETWORK}, perhaps computationally and physiologically skewed, respectively. Finally, these smaller groups are merged into progressively larger groups of users over which RelFbk assessments are pooled.


Top of Page | UP: Other learning applications and issues | ,FOA Home


FOA © R. K. Belew - 00-09-21