Re: [xmca] Improving XMCA as intellectual resource

From: Mike Cole <lchcmike who-is-at>
Date: Fri Jul 04 2008 - 15:00:53 PDT

Thanks Bruce.
Here is a cut and paste description of what the programming team is
considering. If you have a better
idea, let us know! (Also, this is only part of what is being contemplated;
the other part involves connections to external data bases). All ideas, and
especially donations of programming expertise,
greatfully accepted!!


"Generally speaking, it is a searching problem, and we wish to train a
searching algorithm and a classifier (basically the same thing).
Alfredo had a good idea of what this work was and what it entailed.
One important question he raised was how to bootstrap his algorithm --
and he would like us to provide us a list of terms, concepts, words
that appear over and over in the discussion, or that at least his
algortithm should be keyed to as a high-value terms.

On Fri, Jul 4, 2008 at 12:59 PM, BRUCE ROBINSON <> wrote:

> Mike,
> Could you give some more information about the sort of software you are
> going to write and how it would make use of these keywords? Are the keywords
> simply going to be used to tag text in a database? Or is some more complex
> mechanism such as a thesaurus with broader / narrower terms / synonyms being
> considered to aid navigation? A good starting point might be to see if any
> of the existing online databases such as Psychological Abstrscts already use
> one that could provide you with possible terms and structures. (An
> alternative might be to hire an information scientist as a consultant but
> that would cost!)
> As for this problem...
> >Also, there will items that will need further development. For
>> example, the word "object" will need to be
>> tagged in some way to differentiate between "I OBJECT to engaging in this
>> activity" from "The Object of this activity
>> is to improve community intellectual resources" But later for that.
> ...take it from me (I spent a lot of time studying computerised natural
> language processing and writing NL interfaces to databases albeit it was 20
> years ago) that there is no 100% solution through formal analysis of
> syntactic and still less semantic ambiguity. As CHAT-ists would expect, in
> inter-human interaction we make a lot of use of context, what we know about
> the speaker / writer and negotiation and still manage to get it wrong quite
> often. Using a restricted and bounded vocabulary obviously helps but is just
> that - restrictive. The persistence of ambiguity was one of the things that
> helped confirm my suspicion that AI in the full sense really wasn't possible
> precisely because it couldn't provide a substitute for what goes on in our
> everyday social interactions. Useful software artefacts that appear
> intelligent obviously are possible but that's not the same thing. (I've
> always thought the Turing Test was wrong.)
> I don't know how far this is relevant to what you have in mind (and hope it
> doesn't just come over as being pessimistic) but with software projects it
> is always better to err on the side of not being too ambitious as multiple
> disasters in the British public sectore confirm.
> Bruce R
> Here is a quick starter set by way of examples:
>> object
>> activity
>> action
>> operation
>> mediation
>> zpd=zoped=zone of proximal development
>> development
>> learning
>> identity
>> contradiction
>> power
>> gender
>> race
>> ethnicity
>> community
>> play
>> work
>> classroom
>> ........................................................................................................................
>> Apres vous, le deluge
>> mike
>> _______________________________________________
>> xmca mailing list
> --------------------------------------------------------------------------------
> No virus found in this incoming message.
> Checked by AVG.
> Version: 8.0.101 / Virus Database: 270.4.5/1533 - Release Date: 7/3/2008
> 7:19 PM
> _______________________________________________
> xmca mailing list
xmca mailing list
Received on Fri Jul 4 15:01 PDT 2008

This archive was generated by hypermail 2.1.8 : Fri Aug 01 2008 - 00:30:07 PDT