Re: HM.applications-Translations

[Mark Brownell <gizmotron@earthlink.net>]

> I need to understand
> 
> "Each document that is loaded has the option to declare search terms."
> 
> which sounds a lot like an inverted index using what the hytimers 
> called an fcsloc (a numerical location in a coordinate system, sometimes 
> built up by a map from a byte offset).  A numerical system is not 
> heresy to a markup expert, but it may be to some XMLers who are a 
> bit uninformed.  XML shouldn't care.
> 

Hi Len,

Since this application is a functionary that utilizes the devices set in the MTML. 
I would say the declaration of a search item's existence was nothing more than 
a pragmatic solution that lends itself to addressing the problems it would 
experience parsing for the existence of these search items, during the loading 
of each document. For me this is both a speed issue and an inclusion solution. 

It is possible to load MTMLized HTML into the sub-browser without declaring 
the existence of the search items. If this were one document that was part of 
several then only one document needs to declare the search terms for the 
entire group.

Did you notice a verisign addition to the shockwave when you first ran the 
sub-browser. That was the C++ string class being added to shockwave.
I use the string class to facilitate the speed I get out of the numerical 
parsing algorithms. This is so much faster than the Lingo that Director 
uses to do these type of searches. All that is happening here is a parallel 
list of the beginning points of each element's begin & end tags. I use this 
same algorithm over and over whenever it is useful. 

> This
> 
> <foo foo>anything</foo foo>
> 
> should return an illegal character violation.  It is not well-formed XML 
> unless the name production allows whitespace.
> 

I'm telling you, I've just not spent that much time with XML; yet. I'm guessing 
here that white space in XML element signals the parcer to expect an 
attribute to begin. Like <foo foo1="This is the center of foo">anything</foo>

In MTML there is only the consideration for <anything with white space> 
and </anything with white space>. All that is happening here is string 
comparison and whatever is found inside these phrase like handles.

Your questions have forced me to think of other uses for MTML. I was having 
trouble with non-programmers adding MTML to their text articles until I created 
a simple program that allowed them to highlight specific text and then double click 
the correct search item from a separate window in the application. So I came up 
with this idea. MTML can be a simple way for non-programmers to add meaning 
tags to their works. That MTMLized text could then be put through a middleware 
application and transformed into any XML type needed. I like this idea. I'm going 
to do this.

> In a later mail, you mention the meaningfulness of the author in the 
> context of a greater field.   Precisely.  Understand this is why 
> the topic map experts always state a topic map is an "opinion".  
> 

You should see the flame wars at just the suggestion of a construct library 
for a lexicon of Gestalt terms. I'm telling you from experience that if you 
had the perfect machine implemented now, that you would basically have 
very little. The bigger the ego, the higher the boundaries that exist. I have 
done very well to stay out of this and to claim my safe ground as just the 
programmer. So I'm kind of enjoying reading all these ideas floating 
around here in this group. 

> Of great concern to expert system developers (what a closed world 
> model was OUAT), is the issue of authority and legitimacy.   For 
> example, your tool was running in the topic of the gestalt theory. 
> There can be variants of that theory and as a result, one finds 
> oneself having to adopt the view of a school of that topic.  I 
> use the view term deliberately.   Hermaneuticists and semioticians 
> are very careful to state that text analysis depends on a point 
> of view.  It isn't meaningful unless that is accepted.  Once 
> accepted, it is as meaningful as relevant or useful.  If that 
> sounds like politics, it is.  The semantic web is simply a 
> means to remove choices of representation to enable one to 
> discover like and divergent political assertions.  In simplest 
> terms, the choice of the choosers.  All *meaning* is systemic.
> 

That is why I used an author defined model. It is also why 
<tags such as this> 
<I mean top dog> I'm correct, you're not.</I mean top dog> 
</tags such as this> tags such as this work anywhere in the 
text of a MTML document. 

> The semantic web as a service is useful.  As an authority, it 
> is dangerous.  It becomes a golem.
> 

I had to look that one up. I'm working on this MTML with the idea 
that the author of the material is responsible for declaring their on 
inclusion devices for the needs of the greater field. It is to the authors 
own advantage to be inclusive rather than divisive. Although there 
is a Machiavellian advantage to misrepresenting a set of facts in order 
to gain a political advantage. This is carved in stone. 

> One idea of HumanML is to encode cultural systems such that 
> the choice of the golem is in a context relevant to that 
> culture.  This might be considered input to a deontic logic 
> system.   As an expert system, the SW architecture is unremarkable. 
> However, at scale, it can do very serious damage.  For that 
> reason, the power of the Golem to make decisions over some 
> domains is restricted or completely unavailable.
> 
> Len 

The model that I can imagine works like a search engine (yahoo!) that 
puts up a document that is parsed by the enquiring application, for the 
purpose of querying it for the best results desired. It seams to me that 
the search engine needs to receive a set of cultural instructions when it 
is first asked to search. Instead of a keyword or meta-data match I would 
think that there is a greater need to send the search engine a fuller set 
of instructions. In this fashion, I suppose there would be only information 
returned that best fit the set of instructions. I can see the existence of a 
comma delimited search enquiry that is based on string comparisons 
much like the MTML that I'm using now. My RDF solution for MTML has 
comma delimited search terms in its <dc:description> element set.

So this phrase: "my secret fishing spots in the Sierras" would return any 
deposited RDF matches that had this in the <dc:description> element set 
from that search engines RDF search engine. I'm not sure how you would get 
cultural differences to work for this. I mean an equivalent consideration for the 
Sierra Madre in Mexico could also be a cultural fit for this element value. So 
how do you get machines to draw the connection? How about a selection 
option by the user to do a culturally equivalent procedure in the search 
instructions. It will be interesting to see if it can be built so as to not step 
on any toes.

Mark