[open-bibliography] BibSoup/BibServer collaboration model?

Thu Feb 2 18:44:33 UTC 2012

I'm annotating Jim's replies as well.

On Thu, Feb 2, 2012 at 5:54 PM, Jim Pitman <pitman at stat.berkeley.edu> wrote:

> Tom Morris <tfmorris at gmail.com> wrote:
>
> Caveat. Different people care about different aspects of bibliography.
Some people care about the difference between a print version of a
scientific paper and the electronic. Others regard them as the same object.
But given that, Tom's vision is close to mine

>
> > In my perfect world, I'd never have to enter any information which is
> > not unique to me.  Books, authors, journals, articles, author
> > affiliations would all be magically known to the system and all I'd
> > need to add would be my own special value.
>
> My long term goal also.
>
> Yes. We are a lot closer to that vision now than, say, a year ago. I think
we can reasonably envision services that provide much of this. So (real
examples)
* I want to find all the details of my Penguin books. Penguin is a British
company so I expect to find them in the OKD-compliant British National
Bibliography. Mark is getting this installed RSN. Then I can simply search
for publisher=penguin, title="heavy weather" author="wodehouse". As all the
national bibliographies come on line this will be a fantastic resource
* I want to find all papers on malaria. I search medline with all the
keywords (malaria, plasmodium, etc.) Medline has 70,000. Mark/.we are
setting up a bibserver/bibsoup for the OKF open-access group. There are 50
papers per week which the malaria community captures. We can update the
bibserver weekly

The areas that will be more difficult are:
* journal articles outside the mainstream of biomedical , maths, stats,
compsci, physics, economics
* books predating the national bibliographies (I think BNB goes back about
60 years??)
* grey literature (magazines, newspapers, reports, etc.)
* theses (some countries are good, others are not good)

I am also guessing that major providers like Google/Microsoft/Mendeley will
see the value of providing Bibserver-compatible APIs.

> > Reading list?  Just pick the books to be included on the list.
>  Annotated bibliography?  Pick
> > the journal articles and add my comments or category tags of whatever
> > other type of annotation I like.  Reference list? Pick any CSL
> > template and book list and get my reference list generated in the
> > appropriate format to include in my publication.
>
> Exactly.
>
Exactly

>
> > Recognizing that this nirvana is a long way off
>

This depends very much on your field and the type of objects you want.

> > and being willing to
> > help get there, I realize that I'm going to have to enter the data if
> > someone else hasn't already, but I'd like others to benefit from that
> > work and, conversely, be able to reuse the work that they've done to
> > save myself effort.  I know that some of my "data entry" will actually
> > be converting existing resources that I have into a usable form rather
> > than sitting down at the keyboard and retyping things.
>
> Also my view.
>
Any system that helps authors and readers has the opportunity to extract
bibliography in a transparent manner. That's how mendeley get their content
- readers read stuff and provide the metadata

> > Control of my work is important, so I'd like to be able to choose how
> > much, if any, of my annotations, lists, data entry, etc gets shared
> > (or is visible) to others.
>
>
User access/rights can be complex. If you are prepared to do this at a
per-server level then you can use general web technology (login to a site)
without any further work. If you want differential access to records in a
Bibserver I think that's hard, expensive and not worth it at this stage

> > What space, if any, does BibSoup/BibServer occupy in this world?
>
> I think they are a major step towards realizing the goals indicated above.
>
> > If Jim Pitman has a public reading list and I have a reading list, can I
> > easily create a merged reading list with the books from both?
>
> I can, currently, with simple python code operating over BibJSON lists.
> This is not
> yet available as a feature in BibServer, but it could fairly easily be.
> Another feature request.
>

The problem with this is "sameAs" and deduplication. And this may depend
on what the purpose is. If I am worried (say) about different editions then
I need to make the entries distinct. If I don't care then I try to merge
them. This may not be trivial. Identifier systems are very useful here but
not a panacea .

>
> > If Jim's bibliography and my bibliography have a different list of
> > authors for a journal article, is there a collaboration mechanism
> > other than email for coming to agreement, if we both desire, on the
> > correct set of authors and updating the bibliographic database with the
> results of that
> > agreement?
>
> This is a hard social problem, undecidable and we shall still be
discussing this in ten year's time. We have
* identity of authors. Is "Iain Banks" the sameAs "Iain M Banks"?
Biologically yes, in a literary sense  completely not. (see
http://en.wikipedia.org/wiki/Iain_Banks);
* number and order of authors

We are starting to get new deployments almost daily and I think we shall
soon see an O(N^2) level of interest.

>
>
-- 
Peter Murray-Rust
Reader in Molecular Informatics
Unilever Centre, Dep. Of Chemistry
University of Cambridge
CB2 1EW, UK
+44-1223-763069
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.okfn.org/pipermail/open-bibliography/attachments/20120202/9501fff4/attachment-0001.html>