Hi Les,
> Has anyone tried the facilities of base.google.com? It seems
> to be a rudimentary repository - not set up for research
> outputs but for items and events of everyday life. I suppose
> that it's intended to provide an analogue of "Open Access"
> for a much broader range of digital items, and one can see
> how the small but well-chosen sets of metadata will benefit
> Google's other search services.
>
> I did try out the "Reference Article" type and it has a few
> useful fields, so perhaps it could function as a really
> lowlevel self- archiving fallback for independent
> researchers. However, there is no easy way to query Google
> base for research articles, even after you have put them in!
There is a query API described here
http://www.google.com/base/api/demo/html/demo.html#query
I found (a bit to my surprise!) that I have a collection of three items
of type "reference article", which I guess I must have created ages ago
when I was poking around with this.
So I can issue queries like (keyword search, limited to "reference
article" type)
http://www.google.com/base/feeds/snippets/-/reference%20articles/?bq=ope
nurls
Or, again, limited to "reference article" type, on a specific "Field"
(publisher=UKOLN):
http://www.google.com/base/feeds/snippets/-/reference%20articles/?bq=%5B
publisher%3AUKOLN%5D
Result sets are returned as Atom feeds (and I think, though I haven't
checked very hard, they contain all the metadata I put in).
> What I did find very interesting is their approach to
> developing schemas. Whereas others have tried folksonomies,
> that is community- defined sets of keywords, Google Base
> adopts the same idea not to the keywords or attribute values,
> but to the attributes themselves. This if you try to search
> for reference articles about something bland like
> "computers", you get the following search fields "Keywords,
> Pages, Publication volume, Publish date" but if you search
> for "Genes" you get the much richer set "Keywords, Pages,
> Mrna accession, Entrez gene id, Protein accession, Gene,
> Author, Publication name, Csnp, Publication date" because
> gene-related items are frequently described by those attributes.
>
> Perhaps something for the repository platforms to think about?
> --
> Les Carr
Cheers
Pete
---
Pete Johnston
Technical Researcher, Eduserv Foundation
Web: http://www.eduserv.org.uk/foundation/people/petejohnston/
Weblog: http://efoundations.typepad.com/efoundations/
Email: [log in to unmask]
Tel: +44 (0)1225 474323
|