On Sun, Mar 02, 2008 at 08:37:13PM +0000, Hugh Glaser wrote:
+> Sorry.
+> The previous message went out before I finished
+> (Error message - Level 7: brain-finger protocol failure).

:)

+> 
+> On 02/03/2008 19:45, "Joerg Diederich" <[EMAIL PROTECTED]> wrote:
+> 
+> > Hi Hugh,
+> >
+> > On Sun, Mar 02, 2008 at 04:43:20PM +0000, Hugh Glaser wrote:
+> > +> Bad news: We can't link people with the D2R-DBLP, as DBLP (while
+> > generally
+> > +> excellent) has too many different people of the same name with one
+> > URI.
+> >
+> > Yes and No: If Michael Ley (the DBLP maintainer) becomes aware of such
+> > cases, he tries to separate them using _0001, _0002 suffixes to give
+> > people different URIs also (not sure if you know this). So even for my
+> > (in Germany rather not so common) name there is one publication not
+> > being published by me; hence I notified Michael and he renamed the
+> > author to 'Jörg Diederich 0002'. Of course, this is difficult
+> > for Asian people where more people have the same name; in this sense
+> > you are right that there are for sure cases where different people
+> > have the same URI in DBLP (though I have no clue how many there are).
+> Yes, I know, but we can't expect people to sort this out on the scale of
+> 500,000 author entries.
+> It is fine for eyeballing, where a slightly more liberal approach to the
+> problem is appropriate (I want to make sure I see all someone's papers).
+> But for automatic processing, a more conservative approach is appropriate;
+> for example, it would be very dangerous to award research funding based on
+> citation (as the UK is about to do!), where a liberal approach had been
+> used.

Ok.

+> 
+> My grad student, Afraz Jaffri did a study, reported in
+> http://eprints.ecs.soton.ac.uk/15181/
+> Choosing some common names, he got:
+> Name        No. Authors
+> David Smith 15
+> David Williams 10
+> David Jones 8
+> David Evans 7
+> Alan Williams 6
+> Matthew Jones 4
+> Andrew Taylor 4
+> Michael Taylor 4
+> Andrew Brown 4
+> Ben Smith 4

Ah, good to know someone took a look already. 

+> 
+> DBPedia currently has similar problems, by the way.
+> 
+> So what we did was generate a new URI for each author on each paper, and
+> then try to mechanically make conservative decisions on whether two authors
+> were the same person.
+> Our analysis is not yet good enough, but in due course Michael might like to
+> use the analysis to feed back (and I am sure others are doing similar
+> analyses).

In fact, he does a similar kind of analysis (that's why he has the
different colors in the co-author table). Unfortunately this is not
available in the xml-dump of the DBLP data he is providing, so I can't
use it...

Do you use the data in rkbexplorer? I could provide alternative
'author' links (or even replace mine; though I'm not sure how many
people actually rely on the 'original' authors provided in DBLP). 

Also, If you can provide me with the data, I'm happy to take a look
whether I could include it in my database. I guess I'd only need a new
relation dblp_key <-> author (but I'd need a weekly update).

+> In any case, Semantic Web applications having access to both the liberal and
+> conservative treatments is all part of the rich tapestry.

Yes, I wouldn't be using the (Semantic) Web if I expected to get 100%
correct answers... (speaking as a 'liberal' user of the Semantic Web
;-)

+> >
+> > +>
+> > +> Good news: We can link to the papers'URIs, so I have done so.
+> > +>
+> > +> For example:
+> > +> http:°dblp.rkbexplorer.com/id/conf/semweb/PietrigaBKL06
+> > +>
+> > +> So I guess it is now a strong link (363540 links) from
+> > http:°dblp.L3S.de/
+> > +> to http:°dblp.rkbexplorer.com
+> >
+> > And now also in the other direction :)
+> Wow! Excellent.
+> But actually the links should be eg
+> http://dblp.rkbexplorer.com/id/conf/semweb/PietrigaBKL06
+> not
+> http://dblp.rkbexplorer.com/description/conf/semweb/PietrigaBKL06

Ok, fixed. 

Best regards,

/Jörg

+> 
+> Best
+> Hugh
+> >
+> > Best regards,
+> >
+> > /Jörg
+> >
+> >  Sun, Mar 02, 2008 at 04:43:20PM +0000, Hugh Glaser wrote:
+> > +> Bad news: We can't link people with the D2R-DBLP, as DBLP (while 
generally
+> > +> excellent) has too many different people of the same name with one URI.
+> > +>
+> > +> Good news: We can link to the papers' URIs, so I have done so.
+> > +>
+> > +> For example:
+> > +> http://dblp.rkbexplorer.com/id/conf/semweb/PietrigaBKL06
+> > +>
+> > +> So I guess it is now a strong link (363540 links) from 
http://dblp.L3S.de/
+> > +> to http://dblp.rkbexplorer.com
+> > +> There is then linkage to all the other rkbexplorer.com sites.
+> > +> I would suggest that you pick the bigger ones (ACM, Citeseer, CORDIS,
+> > EPSRC,
+> > +> IEEE, NSF, RAE2001, UNLOCODE) explicitly, and the rest in a group.
+> > +> I will think about whether it is safe to link up with ECS Southampton.
+> > +>
+> > +> So according to the picture, it will be possible to get from flickr to
+> > +> DBPedia, and on through DBLPs, and find out what NSF and CORDIS funding
+> > +> someone on flickr has. What fun!
+> > +>
+> > +> Cheers
+> > +> Hugh
+> > +>
+> > +> On 29/2/08 15:10, "Richard Cyganiak" <[EMAIL PROTECTED]> wrote:
+> > +>
+> > +> > Hugh,
+> > +> >
+> > +> > On 28 Feb 2008, at 18:28, Hugh Glaser wrote:
+> > +> >
+> > +> >> Did you consider *.rkbexplorer.com?
+> > +> >
+> > +> > Yes, of course.
+> > +> >
+> > +> > I'm unsure about the extent to which the RKB Explorer dataset is
+> > +> > interlinked with other LOD datasets and wanted to investigate this
+> > +> > before hopefully integrating the dataset into the next update. I need
+> > +> > a couple of links from or to the dataset so I can draw an arrow from
+> > +> > or to an existing bubble.
+> > +> >
+> > +> >> There are about 20 sites there, all with resolvable URIs, SPARQL
+> > +> >> endpoints and CRSes (the service that provides the links).
+> > +> >
+> > +> > Yes, excellent stuff. Linked Data coming out of Southampton is always
+> > +> > top-notch :-)
+> > +> >
+> > +> >> See my foaf file for some links, where I have chosen to use
+> > +> >> owl:sameAs:
+> > +> >> http://users.ecs.soton.ac.uk/hg/foaf.rdf
+> > +> >
+> > +> > Oh, so at least I could draw an arrow from the FOAF bubble to the RKB
+> > +> > one ;-)
+> > +> :-)
+> > +> Actually, that sort of foaf link is the only link for quite a few of the
+> > +> sites, such as ECS Southampton.
+> > +> >
+> > +> > BTW, you got owl:sameAs links to
+> > +> > http://rae2001.rkbexplorer.com/description/resource-Id-227384
+> > +> > and
+> > +> > http://dblp.l3s.de/d2r/page/authors/Hugh_Glaser
+> > +> > both of which are HTML pages about you, but not you.
+> > +> Many thanks; fixed; must stop doing those 303s myself :-)
+> > +> What is the inverse of 303 redirect: 0.003300330033... redirect? Clearly
+> > +> goes on forever.
+> > +> >
+> > +> >> You will see that we are pretty well represented on Sindice now.
+> > +> >
+> > +> > I know ;-)
+> > +> >
+> > +> > By the way, is there an easy way to search for myself in the sites? I
+> > +> > found two identifiers for myself in dblp.rkbexplorer.com, I wonder if
+> > +> > there are more in the other datasets?
+> > +> You can use the "browse" of a repository such as dblp to search for
+> > strings,
+> > +> as you did, and then resolving the URIs gets you to other repositories
+> > where
+> > +> the identification has been made.
+> > +> So resolving a URI such as
+> > +> http://ulm.rkbexplorer.com/id/publications/person-13b99ea8
+> > +> shows the other repositories known about for that URI.
+> > +> Since they are separate reps, there is currently no way of doing a 
string
+> > +> match directly across them all.
+> > +> If you want to search most of them at once, the RKBExplorer itself has a
+> > +> search facility (click "search"):
+> > +> http://www.rkbexplorer.com/explorer/
+> > +> This is showing the two dblp URIs (hover over the links to see the 
proper
+> > +> URI).
+> > +> Berners-Lee is more productive :-)
+> > +> Mind you, we are doing a rebuild this weekend, so things may not be as 
well
+> > +> inter-linked or responsive as you expect.
+> > +> >
+> > +> > Richard
+> > +> >
+> > +> >
+> > +> >>
+> > +> >> Cheers
+> > +> >> Hugh
+> > +> >>
+> > +> >> --
+> > +> >> Hugh Glaser,  Reader
+> > +> >>          Dependable Systems & Software Engineering
+> > +> >>          School of Electronics and Computer Science,
+> > +> >>          University of Southampton,
+> > +> >>          Southampton SO17 1BJ
+> > +> >> Work: +44 (0)23 8059 3670, Fax: +44 (0)23 8059 3045
+> > +> >> Mobile: +44 (0)78 9422 3822, Home: +44 (0)23 8061 5652
+> > +> >> http://www.ecs.soton.ac.uk/~hg/
+> > +> >>
+> > +> >> "If we have a correct theory but merely prate about it, pigeonhole
+> > +> >> it, and do not put it into practice, then the theory, however good,
+> > +> >> is of no significance."
+> > +> >>
+> > +> >>
+> > +> >>
+> > +> >> On 28/2/08 18:14, "Richard Cyganiak" <[EMAIL PROTECTED]> wrote:
+> > +> >>
+> > +> >> All,
+> > +> >>
+> > +> >> http://richard.cyganiak.de/2007/10/lod/
+> > +> >>
+> > +> >> I've made a few additions to the diagram of LOD datasets:
+> > +> >>
+> > +> >> http://dbtune.org/last-fm/
+> > +> >> Yves Raimond's service exports social network data and a list of
+> > +> >> recently listened audio tracks for each user of the AudioScrobbler/
+> > +> >> Last.FM service.
+> > +> >>
+> > +> >> http://qdos.com/
+> > +> >> This Web 2.0 app allows you to measure your digital presence -- how
+> > +> >> much of a footprint do you leave on the Web? My score is 3711, how is
+> > +> >> yours? QDOS has Linked Data output thanks to Steve Harris'
+> > +> >> involvement.
+> > +> >>
+> > +> >>
+> > 
http://apassant.net/blog/2007/12/18/rdf-export-of-flickr-profiles-with-foaf-a
+> > +> >> nd-sioc/
+> > +> >> Alexandre Passant's service exports social network data for Flickr
+> > +> >> users.
+> > +> >>
+> > +> >> http://dbpedia.org/
+> > +> >> The DBpedia team has recently made another major update to the
+> > +> >> dataset.
+> > +> >>
+> > +> >> I'm a user of Flickr and Last.FM, and have signed up to QDOS, so this
+> > +> >> makes three new owl:sameAs links for my FOAF profile :-)
+> > +> >>
+> > +> >> Do you have suggestions for other sites to add to the diagram? I'm
+> > +> >> sure I missed some good stuff ...
+> > +> >>
+> > +> >> Cheers,
+> > +> >> Richard
+> > +> >> _______________________________________________
+> > +> >> Linking-open-data mailing list
+> > +> >> [email protected]
+> > +> >> http://simile.mit.edu/mailman/listinfo/linking-open-data
+> > +> >>
+> > +> >>
+> > +> >>
+> > +> >> _______________________________________________
+> > +> >> Linking-open-data mailing list
+> > +> >> [email protected]
+> > +> >> http://simile.mit.edu/mailman/listinfo/linking-open-data
+> > +> > _______________________________________________
+> > +> > Linking-open-data mailing list
+> > +> > [email protected]
+> > +> > http://simile.mit.edu/mailman/listinfo/linking-open-data
+> > +>
+> > +>
+> > +>
+> > +>
+> > +> _______________________________________________
+> > +> Linking-open-data mailing list
+> > +> [email protected]
+> > +> http://simile.mit.edu/mailman/listinfo/linking-open-data
+> > _______________________________________________
+> > Linking-open-data mailing list
+> > [email protected]
+> > http://simile.mit.edu/mailman/listinfo/linking-open-data
+> >
+> 
+> 
+> _______________________________________________
+> Linking-open-data mailing list
+> [email protected]
+> http://simile.mit.edu/mailman/listinfo/linking-open-data
_______________________________________________
Linking-open-data mailing list
[email protected]
http://simile.mit.edu/mailman/listinfo/linking-open-data

Reply via email to