HPSG translator workflow

[Catalin ]::[ Braescu]

Dear all,

Let me in introduce you myself; my name is Catalin Braescu and I work for
Online Magis Corp., a boutique software house.

We acquired a vary basic English-to-Romanian machine translation software
and we intend to do a complete rewrite of it, using HPSG. The newwer version
(named Pandora Nova) is developing in server-side Java (basically EJB). We
have very good programmers but we lack some principial knowledge of this
domains, so I will humble ask you for a bit of insight.

We are trying to conceptualize the exact workflow within our new translator.

1. Lexical Analyzer converts the initial text into tokens .
2. Parser takes the token stream and builds a syntax tree from it
3. Semantic Checker checks the syntax tree for semantic errors
4. ?


Here are the issues:

A. Can anyone sent me to an online example of a syntax tree for English?
B. Okay, we have a good syntax tree. How will be HPSG be applied to it? What
transformation processes will occur, and in what order?
C. Can anyone describe in plain English how a HPSG translation software is
expected to work, step by step?

Now, please don't flame me. We have some hints about this project but we
fear we do not understand the HPSG translation process corect. This is why I
put everything down to earth in my above questions.

Many thanks in advance for any useful information about how to do HPSG
translation.

Awaiting your next message,


Catalin

--
Catalin Braescu
Online Magus Corp.
http://www.onlinemagus.com