Re: Semantic Bleachers: was Re: Afghanistan War Diary as topic map in Maiana

Hi Patrick,

[Mappish view on DBpedia]
> Ok, so are you suggesting a refinement of the mapping to topic map
> syntax as the next step?

No, you asked for public Topic Maps projects and I suggested DBpedia.
That's all. This discussion is a bit twisted. My initial statement was
and still is: The initial Afghanistan War Diary topic map is a poor
one.

I am not saying that translating the Afghanistan War Diary wouldn't be
a Good Thing, I just answered your call for public Topic Maps
projects. Projects which may be driven by a community where nobody has
time or the knowledge to do something right. As said, translating
DBpedia to Topic Maps is simple.

> I agree there is a low barrier but I am unclear what you are suggesting
> as the next step in having a "mappish" view.

> For example, I was looking at some of the downloadable data sets and saw
> the preview on disambiguation.
> http://downloads.dbpedia.org/preview.php?file=3.5.1_sl_en_sl_disambiguations_en.nt.bz2

> The way I would describe the first 15 uses of Austin would be to say the
> subject has a property with the value "geographic-name," a property with
> the value "Austin," for the US names I would add zip code sets as
> properties, etc.

Maybe you're looking at the wrong source. DBpedia translates also the
Wikipedia disambiguation pages into triples. That's true. It simply
says that the "Austin" may refer to a city in Texas and Arkansas etc.

Each of these "Austin" subjects have their own IRI, of course.

> Actually for all US locations I would add zip code sets as alternative
> identifiers, which will tie in a lot of US government data.

That's also provided by DBpedia. The "disambiguations" dataset simply
converts pages like
<https://secure.wikimedia.org/wikipedia/en/wiki/Austin_%28disambiguation%29>
into triples.

That may be useful or not that useful.

[...]
> PS: I think assigning properties to subjects and then saying on what
> basis we think they are the same as other subjects avoids the pitfalls
> of the OWL sameAs which has no properties nor any way for a user to
> indicate on what basis they claimed "sameAs."

Sure, but if you're referring to the

      <http://dbpedia.org/resource/Alien> <http://dbpedia.org/property/disambiguates> <http://dbpedia.org/resource/Alien_Sun>

triples, you're wrong. DBpedia does not say "Alien" owl:sameAs "Alien
Sun" they invented a property to say that the resource wp:Alien refers
to wp:Alien_Sun and a few other resources.

> PPS: Is this the sort of discussion of "mapness" that you were
> seeking?  

Mapness would: Less dead ends, more associations.

Simple example: If I convert my daily newspaper into a PDF and create
from each paper a topic map like

    newspaper-2010-11-01
       date: 2010-11-01;
       content: <http://example.org/newspaper-2010-11-01.pdf>.

I've won almost nothing. Yes, I can search for newspapers which were
issued at 2010-11-01 but the all the interesting things like articles,
the authors, the subjects etc. are hidden in the occurrence.
This would be a poor example of a topic map.

Best regards,
Lars
-- 
Semagia 
<http://www.semagia.com>

<http://www.topicmaps.de/mailinglist/> German Topic Maps mailing list
<http://tinytim.sourceforge.net/> Open Source Topic Maps engine
<http://mappa.semagia.com/> Mappa - Python Topic Maps engine