Artificial Mind Speech Input Parsing

Thread View: comp.ai.nat-lang

2 messages

[T] Back to Threads [V] Tree View [G] Group Articles

2 total messages Started by Mentifex Thu, 30 Apr 1998 00:00

#95

Author: Mentifex
Date: Thu, 30 Apr 1998 00:00

32 lines
2567 bytes


Let's say that you are processing ASCII verbal input to a robot AI
and you need to make sure that all the parts of speech are assigned
to their proper nodes on a Chomskyan transformational grammar tree.

Would the following technique work?  Instead of forcing the AI
software to "bet everything" on the chosen assumption that the
incoming sentence is of particular transformational pattern, we
could radically simplify the parsing process by letting ALL POSSIBLE
transformational structures try to absorb the sentence, and the
criterion for successful parsing would simply be that the incoming
sentence traverses an entire structure at one pass without aborting:
"If the shoe fits, wear it," so to speak.

If the incoming sentence pattern were successfully to traverse
two widely different sentence structures, AMBIGUITY would be
recognized, e.g.:  "Time flies like an arrow."  Then other factors
in the Forthmind would adopt one interpretation and discard others.

<a href="http://www.scn.org/~mentifex/aisource.html">Mind.forth</a>.
<PRE>
  /^^^^^^^^^^^\ Syntax Strings Together a Thought /^^^^^^^^^^^\
 /visual memory\           ________   semantic   /  auditory   \
|      /--------|-------\ / syntax \  memory    |episodic memory|
|      |  recog-|nition | \________/------------|-------------\ |
|   ___|___     |       |     |flush-vector     |    _______  | |
|  /image  \    |     __|__  / \  _______       |   /stored \ | |
| / percept \   |    /     \/   \/ Verbs \------|--/ phonemes\| |
| \ engrams /---|---/ Nouns \    \_______/      |  \ of words/  |
|  \_______/    |   \_______/-------------------|---\_______/   |
</PRE>

Re: Artificial Mind Speech Input Parsing

#94

Author: Diana Diamond
Date: Fri, 01 May 1998 00:00

41 lines
3026 bytes

The below is a bit like Feynman's approach in physics ... treat everything
as waves as let it all 'happen'...the wave interference then causes
'reality'... 

Chris Lofting
http://www.ozemail.com.au/~ddiamond


Mentifex (mentifex@scn.org) wrote:

: Let's say that you are processing ASCII verbal input to a robot AI
: and you need to make sure that all the parts of speech are assigned
: to their proper nodes on a Chomskyan transformational grammar tree.

: Would the following technique work?  Instead of forcing the AI
: software to "bet everything" on the chosen assumption that the
: incoming sentence is of particular transformational pattern, we
: could radically simplify the parsing process by letting ALL POSSIBLE
: transformational structures try to absorb the sentence, and the
: criterion for successful parsing would simply be that the incoming
: sentence traverses an entire structure at one pass without aborting:
: "If the shoe fits, wear it," so to speak.

: If the incoming sentence pattern were successfully to traverse
: two widely different sentence structures, AMBIGUITY would be
: recognized, e.g.:  "Time flies like an arrow."  Then other factors
: in the Forthmind would adopt one interpretation and discard others.

: <a href="http://www.scn.org/~mentifex/aisource.html">Mind.forth</a>.
: <PRE>
:   /^^^^^^^^^^^\ Syntax Strings Together a Thought /^^^^^^^^^^^\
:  /visual memory\           ________   semantic   /  auditory   \
: |      /--------|-------\ / syntax \  memory    |episodic memory|
: |      |  recog-|nition | \________/------------|-------------\ |
: |   ___|___     |       |     |flush-vector     |    _______  | |
: |  /image  \    |     __|__  / \  _______       |   /stored \ | |
: | / percept \   |    /     \/   \/ Verbs \------|--/ phonemes\| |
: | \ engrams /---|---/ Nouns \    \_______/      |  \ of words/  |
: |  \_______/    |   \_______/-------------------|---\_______/   |
: </PRE>

Thread Navigation

This is a paginated view of messages in the thread with full content displayed inline.

Messages are displayed in chronological order, with the original post highlighted in green.

Use pagination controls to navigate through all messages in large threads.

Back to All Threads