Lists Home |
Date Index |
7/31/2002 4:14:48 AM, "bryan" <email@example.com> wrote:
> we know that google determines result-position by linkage(with
>certain types of linkers being given preference - bloggers for example)
>and that they also attempt to weed-out people attempting to influence
>the results by various unfair or data-skewing methods.
Do they explicitly favor bloggers, or are bloggers' opinions rewarded
because they are prolifigate cross-linkers and update frequently,
forming the kinds of networks that Google looks for?
>Basically it seems to me that way Google has approached the web is as a
>giant problem in Bayesian Analysis, and that this method has been
>relatively successful(at least more successful than other methods have
Hmmm ... then maybe ontologies could help seed the process with "prior
probabilities" or something? Or maybe, let people specify somehow
that the search should be constrained by the sense that words are used
in some vocabulary/ontology , i.e., if I'm looking for information
about "madonna" I mean the religious personage rather than the pop
singer, so I somehow tell Google to use the "christianity" vocabulary
rather then the "pop culture" vocabulary.
Does Google use its human-built Directory structure (which is sortof
an "ontology") in its "free form" searches, or are these separate?
I should get back to work, sigh, but this subject fascinates me.
I heard about SNOMED and the questions that healthcare professionals
would like to use it to answer a couple of years ago. I'd been
thinking of it as a database query problem ... sortof a join
of the clinical data with the vocabulary data. Jonathan Borden
has helped me see that this could also be seen as a "semantic
web", and this thread has made it clear that the question of
how to combine vocabulary/taxonomy/ontology information to
inform web searches or XML queries is wide open for R&D.