RE: [huml] FW: Social Networks based on the plays by Shakespeare

This is the other message that somehow got eaten by my inbox, and sorta makes me wonder what else I have missed. Darn it.

Regardless, this is even more interesting and more what I had in mind in my first response. Having a testbed sandbox as well documented as Shakespeare means that the algorithms mentioned could be very well refined and tested before moving on to using with say, KIF, Knowledge information Format and/or KQML, Knowledge Query and Manipulation Language to do the kinds of processing I mentioned in my first response for developing viable metrics to test our HumanML work with.

The PieSpy dowload was temporarily disabled though I could see the samples when I visited it the first time and was just not found today, but I will try again later.

I'll have to pay better attention to my inbox, and keep checking it against useful tools like the OASIS archives.

* From: "Bullard, Claude L (Len)" <clbullar@ingr.com>

Although speculative, his follow on was quite interesting too.
Quoted below.

len

"There are all sorts of interesting things that one
could do with such software.

BP> What about a diagram of the relations among words
> in a dictionary? (Tony Plate used did something
> like this in Boguraev and Briscoe, Computational
> Lexicography for NLP, but I'm not sure about the
> details of his algorithm.

I mentioned the PieSpy software because it is
open source Java code that could be adapted and
combined with many other kinds of open source
projects.

For starters, you could consider each word in
the dictionary to be a "character" that interacts
with the words in its definition.

For a different perspective on the same words,
you could consider the interactions of words
with other words in their sample sentences.

Once you have the basic program to take
relationships and create diagrams from them,
you can tinker with the algorithms to do
all sorts of tradeoffs. There is a very
large amount of published literature on
these and related subjects, and you can
try mixing and matching different algorthims
and variations of them.

It would also be useful to package parts of
the PieSpy software with other tools that
are also released under the GPL license.

The nice thing about the PieSpy softrware
is that it does the hard part of creating
the graphics. It also provides a way
of creating animations that show how the
relationships evolve as you move through
a text -- or as you move through any kind
of source material from a database, a
knowledge base, a dictionary, the Internet,
or whatever.

John"

From: Rex Brooks [mailto:rexb@starbourne.com]

Thanks, Len,

This is interesting. Might be a good tool to
analyze projected behavior vs. exhibited
behavior. Perhaps, with algernon helping to
produce rules-based applications working with OWL
ontologies and this to track projected v.
exhibited, we can generate some tracking metrics
to gauge how well models perform in recorded
interactions.

Ciao,
Rex

At 2:01 PM -0600 3/11/04, Bullard, Claude L (Len) wrote:
>FYI.
>
>len
>
>-----Original Message-----
>From: John F. Sowa [mailto:sowa@bestweb.net]
>
>A program called PieSpy was designed to monitor social
>networks on Internet Relay Chat (IRC) and draw diagrams
>of who was interacting with whom. Then somebody had
>the bright idea of feeding the plays of Shakespeare
>to the program (since each line of dialog is nicely
>tagged with the speaker's name, just as in IRC).
>
>The result is in a scene-by-scene diagram of
>the social interactions among the characters.
>The attached diagram shows the social network
>for Act II, Scene II, of Antony and Cleopatra.
>
>For more information, complete with animated diagrams
>that show the evolution of the social networks from
>scene to scene, see
>

> http://lister.linux-srv.anlx.net/shakespeare/
> Shakespeare Social Networks
>
>For more information about the PieSpy software
>(written in Java and free for downloads), see

> http://lister.linux-srv.anlx.net/piespy/
> PieSpy - Inferring and Visualizing Social Networks
>
>There are all sorts of other applications that one
>could imagine. With a bit of preprocessing of the
>source data, PieSpy could be used to draw social
>networks from cc lists on email, bibliographies
>of documents, etc.
>
>And the nodes of the diagrams don't have to represent
>people. They could also represent selected terms
>in Peirce's manuscripts -- showing how the different
>terms were related from one document to the next.
>Other applications could include parts of machines,
>concepts in an ontology, or companies in business.
>
>John Sowa
>
>
>
>To unsubscribe from this mailing list (and be
>removed from the roster of the OASIS TC), go to
>http://www.oasis-open.org/apps/org/workgroup/huml/members/leave_workgroup.p

Rex Brooks
GeoAddress: 1361-A Addison, Berkeley, CA, 94702 USA, Earth
W3Address: http://www.starbourne.com
Email: rexb@starbourne.com

huml message