On Feb 4, 2013, at 10:34 AM, Donna Campbell wrote:

> In mentioning "pushing to break down silos more," it brings to mind a
> question I've had about linked data.
> 
> From what I've read thus far, the idea of breaking down silos of
> information seems like a good one in that it makes finding information
> easier but doesn't it also remove some of the markers of finding credible
> sources? Doesn't it blend accurate sources and inaccurate sources?

Yes, yes it does.

The 'intelligence' community has actually been talking about this
problem with RDF for years.  My understanding is that they use
RDF quads (not triples) so that they have an extra parameter to
track the source.  (it might be that they use something larger
than a quad).

From what I remember (the conversation was years ago), they
have to be able to mark information as suspect (eg, they find
that one of the sources is unreliable, then re-run all all of
the analysis without that source's contribution to determine
if they came to the same result).

I don't know enough about the implementation of linked data
systems, so if there's some way to filter which sources are
considered for input, or if there's any tracking of the RDF
triples once they're parsed out.


-Joe