Hello, 

I need to create some complex reports from a database whose data is
structured text (sort of xml or sgml). The reports are generated following a
rather complex set of rules. For example: 

"If text x is present and text y has format f, then apply format z to text
w" 

The text is in a documental database that behaves basically as a binary
sgml/xml document. The code that manipulates it is written in this DBMS's
'scriptiong language', so the text is sort of pre-parsed and I know it's
structure.

The output is always text, transformed following these rules (and there's a
bunch of them and they sometimes depend on other rules). The amount of text
in the database is rather small (approximately 1Mb), it's the
transformations and rules that are a bit complicated. 

I guess this could be considered an forward-chaining expert system. Is Jess
the right tool for this? O:-) 

This system has already been implemented (perpetrated?) with a horrible
scripting language that comes with the dbms used for storing the data, but
it's very hairy and unmaintainable. Working with it is a nightmare and it
must be rewritten. 

I wrote some Python code that generates code in the dbms scripting language.
This is because the so called scripting language is terribly limited and
almost doesn't support any abstraction (no procedures, lousy variables,
lousy loops, etc...). 2000 lines of python code create 3.000.000 LOC in the
lousy 'language'. This huge size (+- 70 Megs of ascii text) creates other
problems and it's really a nightmare. :-( 

After googling a bit about expert systems and Jess, I thought it could be a
much easier and powerful way to implement this kind of system. Is this
correct? 

What books should I read and most importantly, are there any similar open
source projects whose source I can study? O:-) 

Any help will be _greatly_ appreciated. :-) 

Fernando

Reply via email to