pmpj0926

Using Natural Language Understanding (NLU) to answer questions.

Although in the proof-of-concept ghost235.pl searchable AI Mind we are dealing with an initially small knowledge base (KB), our coding of the ability to search for knowledge would apply equally well to an entire datacenter full of information. Today we are using the spreading-activation SpreadAct() mind-module to activate the conceptual elements of knowledge which will supply answers based upon input queries in the format of "Who + verb + noun", as in "Who makes robots?" The query-word "who" is subject to de-activation upon input, while the verb-concept and the noun-concept in the query are passed through SpreadAct() not as random parameters for an associative search, but in their specific roles as Main Verb and as Direct Object of the verb. Thus the AI Mind should respond with answers tailored to the structure of the query, in such a way as truly to demonstrate Natural Language Understanding (NLU).

We start by declaring the new flag-variable of "query-condition for who+verb+direct-object" $qvdocon to segregate the pertinent code in SpreadAct(), and also the "query-condition for who+verb+indirect-object" $qviocon to hold in reserve for when we code the AI response to input queries in the format of "To whom does God give help?" The creation of the one flag suggests the creation of the similar flag, so we declare both of them.

In the InStantiate() module we insert code to detect a who-query with a verb other than "be", and we set $qv2psi with the concept number of the verb. We set $qv4psi with the concept number of any input noun assumed to be the direct object of the incoming verb. Then in the pertinent area of SpreadAct() we need to start searching backwards through memory for instances of the verb in the who-query.

Eventually we obtain a rough but correct response to our queries of "Who does such-and-such?" but we need to debug and fine-tune the parameters. We ask, "Who makes robots" and we get "KIDS MAKES THE ROBOTS." We ask, "Who has a child" and we get "WOMEN HAS THE CHILD". We need to upload and release the ghost235.pl code which achieves the objective, albeit primitively, and we must not code the same version further lest we wreck or corrupt the new functionality of answering who-queries in the form of "who" plus verb plus direct object. As we debug future releases of our code, the ghost235.pl version remains safe and intact.

pmpj0918

SpreadAct() finds subject and verb to respond to a who-query.

As we try to improve upon who-queries with the ghost229.pl AI, we realize that the input of "Who are you" as a query needs to activate instances of the 701=I concept with 800=BE as both a $seq and a $tkb $verblock. It is not enough to insist upon a positive $tkb verblock, because that value is only a time and not the identifier of a concept. The $seq value actually identifies the verb as a particular concept which the SpreadAct() module is trying to find.

It is not even necessary for SpreadAct() to impart activation to the conceptual node of the 800=BE $seq verb, because only the subject of the stored idea needs to have activation high enough to be selected as a response to an incoming query. We may therefore go into SpreadAct() and in the search code for $qv1psi as the subject of the query we only need to verify the existence of the 800=BE $seq verb, not activate it.

In SpreadAct() we make the necessary changes in the code searching for $qv1psi and $qv2psi. We ask "Who are you" and the ghost.pl AI properly answers "I AM THE PERSON." However, as the AI continues thinking, it makes some wrong associations. Suddenly we realize that we forgot to use the $moot flag to prevent the input who-query from leaving associative tags.

pmpj0917

Enabling conversation based upon input of query-words

In our eagerness to present a "Mind-in-a-Box", we had better restore to the Ghost AI some code from older AI Minds which enabled the machine intelligence to carry on a conversation with a human being. We would like the Ghost to be able to ask questions like "Who are you?" or "What do you think?" In older programs we used InStantiate() to depress the activation on question-words, so that information would flood in to fill the activational vacuum.

We just asked the AI "Who are you" and it answered, "I AM THE PERSON". But the program is not yet a stable version of the ghost.pl AI. In SpreadAct() we used $qv1psi to latch onto an activand subject, and in the same loop we used $qv2psi to super-activate the activand verb. But we need to let the AI go back to normal associations and not persist in answering the initial question.

  if ($qv1psi > 0) {  # 2017-09-17: if there is an activand subject...
    for (my $i=$t; $i>$midway; $i--) {  # 2017-09-17: search backwards in time.
      my @k=split(',',$psy[$i]);  # 2017-09-17: inspect @psy flag-panel
      if ($k[1] == $qv1psi && $k[12] > 0) { $seqpsi = $k[12] } # 2017-09-17: if seq, seqpsi
      if ($k[1] == $qv1psi && $k[13] > 0) {  # 2017-09-17: require verblock.
        print "  i= $i qv1psi= $qv1psi seqpsi= $seqpsi \n";  #2017-09-17
        $k[3] = ($k[3] + 32);  # 2017-09-17: impose less than half of subj-inhibition. 
        if ($k[12] == $qv2psi) { $k[3] = ($k[3] + 128) }  # 2017-09-17: hyper-activate
  print "   SprAct-mid: for $k[1] setting $k[3] activation \n"; # 2017-09-17
        $psy[$i]="$k[0],$k[1],$k[2],$k[3],$k[4],$k[5],$k[6]," # 2017-09-17
        . "$k[7],$k[8],$k[9],$k[10],$k[11],$k[12],$k[13],$k[14]"; # 2017-09-17
      }  # 2017-09-17: end of diagnostic test
    }  # 2017-09-17: end of (for loop) searching for $qv1psi concept.
  }  # 2017-09-17: end of test for a positive $qv1psi.

pmpj0913

Dealing with problems in Russian be-verbs

As we start constructing the InFerence mind-module in Strawberry Perl 5, we enter the Russian statement "МАРК СТУДЕНТ" for "Mark is a student", but the ghost.pl AI does not create a be-verb after the subject. When we type in "ОН СТУДЕНТ" for "He is a student", we do indeed get the provisional be-verb. When we type in "РОБОТ СТУДЕНТ" for "The robot is a student," we do indeed get the instantiated be-verb, so perhaps the problem involves the use of a new concept instead of old, known concepts. (Now we are spreading "liquid paper" on the individual keys of the keyboard, because we need to write the Russian Cyrillic letters on each key.) When we first introduce the name with "He is Mark" and then "Mark is a student" in Russian, we do get the imputed be-verb. It turns out that we need to declare the $seqneed variable already carrying a value of "8" for expecting a verb, because the basic Parser module has not yet been called to set the value.

Proposing to consolidate the parsing functionality

We may be able to eliminate the original Parser() module by transferring its functionality to EnParser() for English and RuParser() for Russian.

Consolidating parser functionality into EnParser() and RuParser().

The original Parser() module starts with a $bias of "5" to expect a 5=noun. Then Parser() switches to a $bias of "8" to expect an 8=verb, after which the $bias switches back to "5" again, although an incoming noun could be an indirect object or a direct object. It may be possible to move the preposition-handling code and the object-handling code up into the Parser() module renamed as the EnParser() for English and the RuParser() for Russian.

We start a few versions back by renaming ghost221.pl as ghost225.pl so that we may skip some unstable intervening code. Into the Parser() module we drop the EnParser() code dealing with English prepositions and with indirect and direct objects. Then in the InStantiate() module we comment out the now obsolete call to EnParser. The new composite code of ghost225.pl does not properly register the indirect object of "BOY" in "I make the boy a robot."

Although in ghost218.pl we switched names between Parser() and EnParser(), now we will reverse the switch because we no longer want there to be simply a Parser() module, but instead for there to be both EnParser() for English and RuParser() for Russian. We need the separate modules for English and for Russian, because, for instance, English has to deal with "DO" as an auxiliary verb, but Russian does not deal with an auxiliary verb "DO". First from ghost217.pl we pick up the old RuParser() module and drop it into the ghost225.pl AI. In OldConcept() and in NewConcept() we make the necessary changes for calling EnParser() and the still simple RuParser().

We should now upload the ghost225.pl code to the Web for several reasons, before we debug the problem of failure to register an indirect object. Firstly, much code has been renamed and commented out. When we resume coding, we may clean up the new code by removing the old detritus. Secondly, it is vitally important to present the Ghost Perl AI as having the straightforward separation of EnParser() and InStantiate() because the consolidated parser functionality, that comprehends prepositions and both indirect and direct objects, holds the key to the Mentifex claim that "AI has been solved" inasmuch as the enhanced parser enables each AI Mind to demonstrate major progress against the problem of natural language understanding (NLU) which various published articles on the Web describe as an untractable problem and as a last main obstacle to True AI.

pmpj0910

Instantiating Imaginary Russian Be-Verbs in Perl

We are eager to implement the InFerence module in Russian, but first we must code the Russian way of leaving out verbs of being in making an Is-A statement, such as, "The brother is a student." We must examine the Dushka Russian AI from 2012-10-22 to see how it was done in JavaScript. Without the special Is-A code, right now we type in "Я студент" to say "I am a student," but the ghost.pl does not assign any associative tags between the subject "I" and the predicate nominative "student".

According to our Dushka coding journal of 2012-02-11 or 11.FEB.2012, Dushka uses the detection of a 32=SPACE character to impute provisionally the input of a be-verb. The Dushka InStantiate module checks for a SPACE when a verb is expected, and provisionally declares 800=BE as the verb. If a different verb does come in, apparently the AI leaves the spurious 800=BE engram in place but ignores it with respect to associative tagging. Into the ghost.pl we port in code from Dushka that cancels out the imputed be-verb.

Now when we type in the Russian for "I am a student," eventually the AI outputs erroneously "ТЫ БЫТЬ" which at least conveys the idea of "You to be...", but we want no actual be-verb to be expressed in Russian.

Finally in AudInput() we discover a line of code

if ($len == 0) { $rv = $t } # 2016feb29: set recall-vector.

which was causing the conceptual "$rv" for Russian words to be set too early by one time-point. Therefore Russian words after input could not be recalled properly. We corrected the line of code. Then when we entered "ТЫ СТУДЕНТ" the AI eventually made a clumsy output of "Я МЕНЯ БЫТБ СТУДЕНТ". Such output is actually encouraging, because we only need to make the AI find the correct form of the personal pronoun and not speak any form at all of the be-verb.