[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lojban] Semantic Representation of Lojban



Hi Alan,

Thanks for your response and the crowdsourcing idea, and for linking to Jorne. As you suspected, I did only receive one or two inquiries about the linking effort.

I like your idea of splitting a Wordnet linking project into specific chunks, each one centered around a group of related concepts. I discovered a very old mapping of gismu to Roget codes, which could be a starting point. I think I need to make this process super easy and painless if I expect anyone to get involved.

What if I created some web pages that would allow Lojban experts to pick a gismu, then select related Wordnet or dbpedia entries from a list, and automatically save it to a database?

We could then make the jorne web service accept Lojban text, and respond with related references from the Linked Data cloud. With this, you can follow that data to English and other Wordnets, Library of Congress subject headings, wikipedia/dbpedia, and the rest of the Linked Data "cloud".

The Semantic Web conference was fantastic! It was overwhelming, but in a "too many sessions and not enough time" sort of way. The latest tools are amazing, and adoption is reaching a critical mass. I am considering submitting a talk for Semtech 2011 on why Lojban would work well with the semantic web. The due date is Monday, so I better get cracking on it ASAP. Wish me luck!

As far as source code for my simple parse tree service, it's a real hack right now, but I will clean it up a little and put it on the jorne sourceforge.

Last year I had intended to get more involved in this, but ran out of time to work on it. Now that there seems to be some interest from both you and Oren, and others, it's time to make this a reality.

Are you offering to own the farm animal words? (asking rather sheepishly...) ;-)

Regards,
Brian Eubanks (iuban)


Quoting ".alyn.post." <alyn.post@lodockikumazvati.org>:

On Thu, Jan 06, 2011 at 07:09:25PM -0500, Brian D. Eubanks wrote:

This is related to what I've wanted to do with the jorne.org project.
I am definitely interested in being a part of the discussion. No time
to go to Penguicon, but I am willing to chat with the group.

Reports of Semantic Web's death have been greatly exaggerated, and I
can assure you that it is quite alive and well. Early adopters include
Best Buy, US Library of Congress, Reuters, US Census, US DoD, EU
governments, Google (Rich Snippets), and biomedical researchers, to
mention only a few who are using and publishing Linked Data and are
willing to put serious time and money into it. Perhaps Robin meant
something different by his use of "Semantic Web", but a large part of
Tim Berners-Lee's original vision is being implemented now in a big way.

Lojban is unique among languages in that the entire Lojban corpus
could become a part of the Linked Data "cloud" (see linkeddata.org).
To make this happen, you would:

1. parse text into predicates (moderately hard but doable)
2. convert predicates to RDF (easy enough once you have the
predicates, mostly this involves defining a standard URI/ID for each
primitive)
3. publish RDF/OWL correlations between each Lojban primitive and
dbpedia, wordnet, and other linked data sets (this is a manual process
done once for each Lojban primitive).

I posted something about this before, but I do have a very rudimentary
web service that takes Lojban text and returns the parse tree as XML
(using Robin's PEG parser). I haven't had time to work much on it, but
I started to look at using Jena (a Java RDF API) to produce RDF as
output instead. I would need help with converting the tree into
predicates, since my Lojban ability is limited.

See
http://groups.google.com/group/lojban/browse_thread/thread/b39f94b183cf344f
and
http://groups.google.com/group/lojban/browse_thread/thread/7e48f11c43f62b63/68095f3ae30b02dd

The web service is at http://jorne.org/form.jsp
Try pasting in some short text examples and let me know what you think.


You know, I remember now seeing this project the first time you
posted about it, I'm not sure why I didn't say anything at the time.

I've added a link to your project on the sidebar to my website.  See
roughly in the middle of "Lojban Resources":

  http://lodockikumazvati.org/

One of our summer researchers recently gave a presentation on the
semantic web, focusing on DBpedia[1], which extracts formatted
information from wikipedia (like sidebar tables) and allows
structured access to that data (using SPARQL and I'm sure other
query methods).  [Brian of course knows about DBpedia and the
related technology already.]

That was a pretty fascinating talk, In that "We could use two or
three of these features right now to make our product better" kind
of way.

I note, reading your referenced threads, that you said:

  Anyone with any background in RDF or Lojban is welcome to join the
  project. I particularly need help with mapping the Lojban gismu to
  synsets in English and other Wordnets. That would bring immediate
  benefit in being able to generate simple lojban sentences from
  dbpedia  and other linked data content. This part could done without
  any  knowledge in RDF. Drop me a line if you are interested in mapping
  some gismu.

I'd love to be wrong, but I'd wager you didn't get any responses to
this request.  I think it is too large, uninteresting, or abstract.
I wonder what would happen if you said "I'd love it if one of you
could take all the gismu referring to farm animals and give me the
link X and Y from dataset Z for them."  Substituting farm animals
for whatever topic seems reasonably interesting and repeating as
necessary.  The idea being to make a focused, specific request that
someone could just go do.  There are definitely subjects for which
I'm a close enough domain expert to do this work, while for others
I'd just rather someone else with more motivation work on them.

Are you interested in experimenting with this kind of crowdsourcing
to build this data?  Is that your next step?  Oren showed a heck of
a lot of interest in your project, it took a non-zero amount of time
to give such a thoughtful reply!

How was the Semantic Web conference in June?  Do we get to see the
source code?

-Alan

1: http://dbpedia.org/
--
.i ko djuno fi le do sevzi

--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com. For more options, visit this group at http://groups.google.com/group/lojban?hl=en.



--
You received this message because you are subscribed to the Google Groups "lojban" group.
To post to this group, send email to lojban@googlegroups.com.
To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
For more options, visit this group at http://groups.google.com/group/lojban?hl=en.