Wed.15.SEP.2010 -- Progress on Predicate
In frustration we want to know one thing. Why can't the 14sep10A.F MindForth answer "I AM ANDRU" when asked immediately "what are you", but it can indeed state that answer if the user first makes some different input, like the word "I"?
In December of 2009 and January of 2010, when we
were coding who-queries and what-queries, we were not yet
using neural inhibition, and so we may have made our code
overly complicated. For example, on about 19jan2010 we
used a SubjectClear module to zero out subject concepts
after the input of a query. If we rely on InHibition
instead, we may get an immediate answer to a query, after
which the answer-idea goes briefly into InHibition and
lets one or more additional answers be thought up by the
All right, fellow cyborgs. We went into the VerbPhrase module and we commented out the InHibit code both before and after the call to SpeechAct, because the pre-SpeechAct code inhibits the idea that has just been resurrected from the knowledge base (KB), and the post-SpeechAct code inhibits the new idea that has just been uttered. Immediately (although we do not know why) our AI began to function better in the answering of "What are you?" queries. We had the following conversation with the MindForth AI Mind.
Transcript of AI Mind interview at 17 48 13 o'clock on 15 September 2010.
Human: you are a program Robot: PROGRAM IS PROGRAM
Human: what are you Robot: PROGRAM
Human: what are you Robot: I AM ANDRU
Human: what are you Robot: I AM PROGRAM
A ".psi" report indicates that only nouns were being inhibited, not verbs.
Btw (nerdspeak for "by the way"), we just enhanced our AI program considerably with a little trick. We did not like seeing the AI say, "I AM PROGRAM", so we decided to reinstate a call from the VerbPhrase mind-module to the EnArticle (English article) module. This time, however, we inserted the EnArticle call just before the main VerbPhrase call to SpeechAct, and we set up IF-THEN conditionality to call EnArticle only for a predicate nominative. Immediately the AI began saying "I AM AN ANDRU" and "I AM A PROGRAM". Somehow we feel that this same enhancement brings us closer to Is-a considerations and to the logical treatment of sets and classes.