LEXTENDER - Project Release Notes

This specs recent and planned progress on LEXTENDER, which is distributed among:
  1. com/lexikos/utility - a package that now compiles into root classes
  2. psi/readme.htm - documenting TODO on machinery.txt and concepts.txt includes
  3. psi/psifolder.htm -  (outdated) specs on how to package lexikons into folders

on 8/22 I copied old core.ltm directly into omnigator's WEBINF/topicmaps/ folder and played
with it to learn about fragility of LTM - 'doc' was unnamed if used before being declared (?)

I also learned the whole package here is VERY confused, and #1 code needs refactoring
Fixing this will be a non-trivial task - should also think about new Eclipse, and Ant (Bowdoin)

When one runs the index.jsp file in lextend/situation folder per [3], it includes [2], which call [1]. 

That is Lextender's pattern, which this /lextend/ folder (was csdemo) will debug.  Then its
logic can be reused almost at once in the psi/ctm/situation folder, to be released!

Old code, which uses a now-patched initializer and psifiles/_index.htm is still able to work
because the index.jsp it uses was pathed to add .txt to the filename in the above


Recent changes to the whole lextender system are

Basic Features prepare for this MODELER upgrade

Q3 era -let WORDS scripts be added (or posted) into Lextender, as well as PSI-trees

They still generate LTM output, but now it comes from an English-like input syntax.
The core processor here will be JScheme, and it will speed up our R&D progress
WORDS scripts, for security, will be embedded in the Scheme language

Both forms of scripts enable more powerful forms of data conversion into TMs,
such as pulling raw data from a DB or mining a normal bookmark file for URIs.
Users can do some things; we can do even more, and add many library utilities .
 
Q4 era - let WORDS scripts and their LTM interact with Context, a new TM engine

This (TM4J) engine is open source, and scripts can mine or control it at fairly low
levels of detail.   This is a major enhancement - we can add utilities that can make
decisions based on what is already in loaded in this TM engine. 

Such utilities can let us emit LTM charts that turn into a contextual discourse model
when loaded, and that discourse model can in turn support anaphora resolution.

TM4J does NOT have a built in presentation web app at this time, but it does have
a set of Java tools for building UIs based on various tools like Velocity.


Below are notes from first Miles era to remove later....

Q1 '04: Original demo based on LTM and Omnigator 7 (a debug aid)

The first link above shows what I built for Miles, which created the LTM
The second link shows the presentation tool he used for it - worked great!

The main ontology I delivered was refined (by hand) to populate instances
See nycplan for plans of original demo ontology and its possible furtures.

Q2 '04: Another demo, now meant to combine the above with PSI doc

On June 29, I played around with code in B and hit a wall - after changing
_lexikon.txt to foo.txt, abnd adding debug printlns, I could not get the second
JSP to pass the right file name in a way the code could accept.  Don't know why
but I will print this out, and book mark it, and pick it up later (in Q3)

Also noted June 29 I  did not grok part A, below

A) This sub-goal requires a new lexikos.utility release to add Concepts.java
B) This sub-goal also requires new psi-lex-folder management conventions for MFCS