Newman is a project mainly written in Python, based on the Apache-2.0 license.
My CS-572 project (Natural English With Mutating Abridged Nouns)
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:
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)