I'm a solutions architect at Innodata Isogen, where I've been involved in publishing process design, DITA, ebooks, and general XML-related architecture work. My work email address is bducharme@ their domain name.
I have a BA in religion from Columbia University and received my
masters in computer science from
New York University's Courant
Institute in the spring of 1996.
Three papers I wrote in pursuit of my masters (note that they have an
outdated e-mail address under their titles in the Acrobat versions):
I wrote "Object-Oriented Development of SGML Applications" (48K Acrobat file) as part of an independent study
project in the spring of 1993 to analyze the benefits that current OO
literature could offer to SGML developers. Although the general ideas
about OO-SGML correspondence in the paper have been superceded by the
tremendous work done on groves and
property sets and the Document
Object Model since then, I'm still proud of the Smalltalk program
that I wrote to demonstrate the paper's ideas. It reads a DTD,
declares classes for each element type, and then reads the output of
James Clark's nsgmls, instantiating
each element to the appropriate class. It also reads formatting
scripts to convert instantiated objects of simple HTML and
TEI documents to RTF and TeX.
The paper was aimed at people (for example, project adviser
Professor Ed Schonberg) who knew more about OO technology than about
SGML. I presented a shorter, revised version of the paper for people
who knew more about SGML than OO technology at SGML '95 in Boston, and
the QUE Book "Using SGML" included a much dumbed-down version of the
SGML '95 paper as its last chapter. Talk about repurposing your content.
In the fall semester of 1993, I wrote the paper "Increasing
Concurrency in Object-Oriented Databases Using Semantic Information: A
Survey" (64K Acrobat file) for Dennis
Shasha's "Advanced Databases" course. This paper analyzed three others
that explored ways to use the semantic information stored in an OO
database to minimize the situations where one process must be held up
until another is finished. (An RDBMS example would be blocking the
update of a particular record until another update of that record had
completed.) As per the assignment, the paper includes four discussion
questions and answers at the end.
In the spring of 1996, I submitted the paper "solfish: Generating
Text from an Existing Grammar" (17K Acrobat
file) for Ralph Grishman's "Natural Language Processing"
course. It used an extensive natural language parsing system written
at NYU in Allegro Common Lisp with a domain of course registration, so
that you could enter English assertions and queries about students and
course registration. I wrote a LISP program to randomly generate
sentences from this grammar with the stated purpose of calling
attention to weaknesses in the encoded grammar by generating sentences
considered correct by that grammar but incorrect by a human
proofreader. The other purpose was just for fun--to see what kinds of
sentences it would generate, especially after the system's vocabulary
was augmented with some more poetic vocabulary entries. (Sample
output: "Her hazy ghosts only eat with fish. Ghosts fish often with
moons. Fish mean to think. Zombie eggs generally sleep.")
See The
Postmodernism Generator (start with the notes at the bottom before
you try to read the actual essay you see) for a more recent, much more
sophisticated application of the same principle.
![[XSLT Quickly cover]](img/XQcover_t.jpg)
![[Ann. Spec book cover]](img/xmlassmall.gif)
![[SGML CD book cover]](img/sgmlcdsmall.jpg)
![[QUE book cover]](img/quebook.gif)
