Pythia

This page is currently in development as of February 12, 2013.

Pythia is my natural language research project to develop a Prolog implementation of a wide-coverage English language grammar. The project is designed to parse and convert English sentences into a standard logical form from which application command based decisions can be made.

Phase 1.  Develop a sufficiently wide-coverage grammar for English  (70% complete)

Phase 2.  Develop a logical form representation for semantic verification of the syntactic parse.   (10% complete)

Phase 3.  Develop a deductive reasoning system for contextual evaluation of natural language statements.  (0% complete)

What does the Pythia – The Knowledge Oracle do?

The parser is built to use a feature grammar loosely based upon the work of Allen, Natural Language Understanding, 2nd edition, ISBN 0-8053-0334-0.  

The goal of this work is to develop a natural language interface to a knowledge system. People have different degrees of knowledge in various domains. Given that a domain of knowledge can be defined, learning can be achieved by asking questions.

The NLP Parser uses a Java graphical interface to access an SWI-Prolog application system. The Prolog system can access the WordNet 2.0 database of over 200,000 words.

pythia1a

pythia1b