[openbiblio-dev] BibServer and approaching contacts

Jim Pitman pitman at stat.Berkeley.EDU
Thu Feb 2 18:48:58 UTC 2012


Mark MacGillivray <mark at cottagelabs.com> wrote:

> From bibsonomy for example we can already pull bibtex. 

Is there an examplar of this? I think there are issues with bibsonomy paging
of bibtex collections which I have not seen addressed. I think this requires
a fairly simple parser/connector e.g. to pull all the records in a collection like
http://www.bibsonomy.org/user/pitman
This collection has 2815 bookmarks and 528 bibtex items. I do not know how to 
push either of these collections to bibsoup.

> Thus people can use bibserver for their collection, which is good because individuals are our
> target. 

Strong supoprt for that. We should be able to provide push button upload and refresh on BibSoup
for any user collection on BibSonomy, both bibtex and bookmarks. I suggest this as a milestone. 
We are not there yet, but it should not be far down the road.

> Thus our approach to these orgs need only be to ask if they would like to output bibjson and 
> if they would like to sign the open biblio principles.

OK. What about the issue of whether a third  party could push BibSonomy data to BibSoup?

e.g. http://www.bibsonomy.org/search/regier
is an interesting collection curated by a student of mine. Whose permission, if anyone's is needed
before such a collection can be pushed to BibSoup? 
This brings back the question of licensing of items on BibSoup. We talked before about
a license agreement on upload. But what if people are uploading from 3rd party sites?

This sort of issue would be simpler if BibSonomy agreed on an explicit
 license for all of its data, which I dont think it currently has. DBLP is much simpler,
because it has declared an open license.

> Regarding the question of scale, yes we could do it, but why? 

Because it would deal once and for all with the licensing issue. e.g. if all of DBLP
is embedded in BibSoup with an explicit licence , then users know exactly where they stand:
they are free to collect and annotate records however they like. Same for BibSonomy, if they
will agree to the bulk transfer.  If not, then it is pushed out to the user to tell they are 
allowed and not allowed to do with data from various sources. We could also try some sort of
education program about what can be done with data from various sources.

> Our interest is the individual or small group. So the only question is helping them get
> their data into bibjson, then they can build and manage their own collection. 

Yes. I strongly supoprt that. But for me as an individual, my biblio capabilities would
be enormously enhanced if all of some large biblio stores were kept updated in BibSoup
(or other similar BibServer instance), especially for me arXiv and for others BibSonomy, DBLP, 
RePEc, ....
So I see a big value for individuals in embedding whatever large stores can be accomodated
in current BibSoup or conmparable BibServer instance.

> Anyone that wants to run a bibserver containing all of dblp is
> already free to do so. That should be dblp themselves, or an interested third party.

Right, and I think we should encourage this, not just for DBLP, but for all the major
open and semi-open biblio stores. How best can we assist that with available resources?

--Jim


> On Feb 2, 2012 5:57 PM, "Peter Murray-Rust" <pm286 at cam.ac.uk> wrote:
>
> >
> >
> > On Thu, Feb 2, 2012 at 5:45 PM, Naomi Lillie <naomi.lillie at okfn.org>wrote:
> >
> >> All,
> >>
> >> Please see below from Jim which follows the discussion from Wednesday's
> >> stand-up - see line 15 of http://openbiblio.okfnpad.org/catchup - all
> >> comments welcome:
> >>
> >> We discussed some of this after Jim's machine dropped... Please correct
> > if I get it wrong.
> >
> >
> >> Before approaching BibSonomy, DBLP, or any other major biblio data
> >> provider, we should confirm between us what we are asking, and have a
> >> roadmap for technical communication and a clear sequence of steps to
> >> proceed. We should have a fairly standard email communication, which e.g. I
> >> or someone else with previous contact could send the right person, and
> >> there is a clear sequence of actions that person can take, and responses
> >> that can be expected on our side. Strategic issues are: do we write a
> >> parser first and show them a demo?
> >> I am concerned about raising expectations of our performance until we are
> >> sure we can deliver. Can we embed all of BibSonomy or DBLP in BibSoup?
> >> Would we want to do that?
> >
> >
> > I think we limited this in the first instance to approaching Bibsonomy
> > about adding BibJSON output.
> >
> >
> >> Or do we only want to embed selected subsets? Or allow users to curate
> >> selected subsets? This issue arises generically with any large source.
> >
> >
> > If there is an Open licence then the issues is primarily of being good
> > citizens (e.g. responsible downloading and giving credit). If not, then the
> > licence issues are the most immediate ones
> >
> >
> >> With DBLP we have an open license already, so I think we should demo
> >> capability before asking for anything more.
> >
> > Agreed.
> >
> >> I already wrote a crude DBLP parser, and this could easily be
> >> improved/installed. We should have a plan for these things, before we open
> >> up negotiations. I think our strategic aim should be to recruit DBLP and
> >> BibSonomy as early nodes of the BKN which will  provide proof of concept
> >> and demo the value of BibJSON and BibServer installations over BibJSON. I'd
> >> like to hear what Mark thinks about the technical feasibility of our
> >> operating on this scale before continuing.
> >>
> >
> > Sound reasonable. My view is that we should have something new and
> > complementary to offer. Some sites may welcome additional means of
> > distribution, while others may see this as an extra burden.
> >
> >>
> >> > secondly, re Rufus's discussion points and your following comments, 'a)
> >> installation of at least one public facing bibserver on a machine not
> >> controlled by cottagelabs/OKF', please could you follow up with people @
> >> Berkeley about getting a BibServer installed there, as this is an easy win
> >> if that can be sorted!
> >>
> >> This is easier said than done. I can keep pushing on this, but I cannot
> >> promise you a time line. My own personal interest puts higher priority on a
> >> SaS arrangement with Mark for the Probability Web. I am sick and tired of
> >> pleading with sys admins and other dept staff to provide adequate support.
> >> Peter is familiar with the problem.
> >
> >
> > We shouldn't push too much at closed doors. I think there will soon be a
> > number of opportunities. I am talking with Open publishers (in the most
> > general sense) and we are mutually keen to explore what this type of
> > technology can offer.
> >
> >
> >> I just want a BibServer that I can easily manage and control and that
> >> will serve the communities I care about. I continue to hope Mark can
> >> provide that on a SaS basis. I think Mark you need to be looking at other
> >> places for installs so you get direct feedback with about what the
> >> obstacles are to be overcome. I do have a number of possible places we
> >> could consider approaching.
> >>
> >
> > My understanding is that it is possible to run this from the cloud? What
> > are the issues involved?
> > * Money.
> > * updates
> > * customisation
> >
> > These are all addressable when there is funding - we need to get an idea
> > of scale.
> >
> >   But there will be some parsing/scripting work to be done.
> >
> >
> > Yes. I have just looked at the Pubmed RIS. It needs work.
> >
> >
> >> Probably we should proceed in parallel on these and others. But I am
> >> unclear what the steps are, because we have not even done one yet.
> >>
> >
> > the Pubmed is a good place to start.
> >
> >
> >>
> >> --
> > Peter Murray-Rust
> > Reader in Molecular Informatics
> > Unilever Centre, Dep. Of Chemistry
> > University of Cambridge
> > CB2 1EW, UK
> > +44-1223-763069
> >
> > _______________________________________________
> > openbiblio-dev mailing list
> > openbiblio-dev at lists.okfn.org
> > http://lists.okfn.org/mailman/listinfo/openbiblio-dev
> >
> >




More information about the openbiblio-dev mailing list