NEWMAN:
Natural English With Mutating Abridged Nouns
Written for CS-572 at UCCS (Computational Linguistics)
By Chris Eberle <[email protected]>

Goal:

Translate natural english queries into a set of predefined
classifier switches. For example, "A man outside with glasses"
would be transformed into "Male => Yes, Outside => Yes,
Eyeglasses => Yes", assuming that "Male", "Outside", and
"Sunglasses" were defined as valid classifiers.

A richer vocabulary is supported with the help of WordNet
which takes unknown words and maps them to known similar
words. The translated sentence is then input into a CFG
which produces output symbols (MALE, OUTSIDE, GLASSES).
These are then used to output the final classifier names
and values.

Prerequisites:

* Python 2.6
* NLTK 2.0

Configuring:

NEWMAN can be extended to support any vocabulary, grammar, and
output. You need to change config.py. There are a lot of
comments in there already, go have a peek.

Running:

$> python newman.py -s "grinning chinese dude"
(Asian => Yes, Male => Yes, Smiling => Yes)