Hi,
Of course it will be full of inaccuracies and problems; this is just a
'labs' project and the first iteration.
I think the interesting thing though is that the results are actually fairly
good (considering the range of data it's tackling, and the primitiveness of
the technology), and will undoubtedly get better quite quickly.
It may never reach the near-100% accuracy of data that official
cross-organisational projects try so hard to deliver, but could it become
'good enough' for most (non-academic/research) uses?
Google have always talked about avoiding explicit 'semantic' data, and have
concentrated on extracting data from normal text. This seems like an
interesting output from such an approach, and can only get better as the
algorithms behind the parsing of text improve, and could even extract data
(such as the emotional context) that we haven't even thought about sharing
(e.g. How many people LIKE or HATE a painting, not just the raw database
facts and figures).
On 04/06/2009 11:08, "Joseph Padfield" <[log in to unmask]>
wrote:
> Hi,
>
> I tried the da vinci search:
> http://www.google.com/squared/search?q=da+vinci+paintings
>
> Then I added a new column: "location", and the results where very suspect.
>
> Semantic information is wonderful, but bad semantics is dangerous. It
> seems like a great service, but some degree of editing and checking
> would be needed. I guess until museums are exporting trusted semantic
> data, this sort of service will need to be classed as "of possible
> interest" but suspect.
>
> Joe Padfield
****************************************************************
For mcg information and to manage your subscription to the list,
visit the website at http://www.museumscomputergroup.org.uk
****************************************************************
|