| Brainhat programming
centers about a collection of definitions known as the Basic
Knowledge Pool. The Basic Knowledge Pool describes the words
and concepts that Brainhat knows at start-up, and details their
relationships to one another. The lineage that makes a ball
a toy, and makes a toy a thing, for
instance, can be traced back to the basic knowledge pool.
In addition to the Basic
Knowledge Pool, Brainhat also depends upon propositions to
motivate inference or cause-and-effect. These are not
programmed into the basic knowledge pool. Instead, Brainhat learns
these through English statements, once the program has
bootstrapped (see Building
Brainhat Scenarios). You might tell the program "if
thing1 is near thing2 then thing2 is near thing1," for
instance.
In addition to concept
definitions, the Basic Knowledge Pool contains rules that tell
Brainhat how to parse input and how to recognize ideas floating
about in core. Rule definition isn't covered in this document.
Fortunately, you shouldn't have to fool with parsing rules in
order to create a vocabulary to match your application.
Concept definitions can be found
in the data subdirectory.
Concept
Definitions
-
Concept definitions can
appear in any order. One definition ends and another begins
whenever Brainhat encounters a define statement. The
name given to the definition must be unique. By convention, I
have been cardinally ordering different definitions of the
same word, e.g. ball-1, ball-2, etc., where
the first might describe a toy, the second a grand social
event.
define ball-1
label ball
child toy-1
wants color-1
wants size-1
typically round-1
The block above shows some of
the basic elements of a concept definition. Statements
subsequent to define can appear in any order. Tab
characters position the columns. Most other characters count,
so when defining concepts, take care to avoid trailing blanks.
- label
- Concepts can have one or
more labels, each representing a synonym for the concept.
The labels can be from one to three words long.
define ball-1
label ball
child toy-1
wants color-1
wants size-1
typically round-1
The definition above
shows the use of a wants tag. As with child
tags, the rightmost field lists a concept name, not a
label.
- typically
- Like a wants tag,
a typically flag specifies a feature of the
object being defined. The difference is that satisfying a typically
flag scores lower than satisfying a wants tag.
- orthogonal
- Two concepts are
"orthogonal" when they cannot exist together. If
you require that something cannot be both red and blue at
the same, for instance, then you can declare concepts
"red" and "blue" to be orthogonal
with respect to "color."
define blue-1
label blue
child color-1
orthogonal color-1
define red-1
label red
child color-1
orthogonal color-1
define pink-1
label pink
child red-1
orthogonal color-1
Any concept can serve as
the basis for the orthogonality of others. And a given
concept can be orthogonal along many axes--color, size,
whatever. Note too that orthogonality takes lineage into
account. You may say, for example, say that children of red-1,
such as pink-1, are orthogonal to color-1,
just as red-1 is. But this will not imply that pink-1
is orthogonal to red-1.
- related
- A related tag
helps Brainhat score a concept against other related
concepts already committed to the context buffer. An
example probably explains this best:
define fun-1
label fun
child things
related happy-1
A previous reference to a
"happy" in the context buffer will help the
above definition of "fun" score favorably. Note
that related tag is different than the wants
and typically tags we looked at previously: the
concept nominated by a related tag does not need
to appear in direct combination "fun." Rather,
"fun" would have to be part of the on-going
context.
- tense
- The tense tag is
primarily for use with verbs and auxiliary verbs
("enablers") (can, will, does, did). Possible
values include past, present, future
imperfect, and so-on).
- number
- The number tag is
commonly used with verbs and nouns. Possible values
include single and plural. Other
possibilities will follow as Brainhat grows up.
- person
- Used primarily with verbs,
the person tag can take the values first,
second and third.
Definition
Guidelines
-
Definitions for a few types
of speech should follow conventions spelled out in this
section. This will allow Brainhat relate specific forms back
to more general forms.
- verbs
- All verbs must be able to
trace their lineage back to action. This means
that, by following child links, you should be
able to find action as an ancestor.
Verbs should be organized
so that the infinitive form is a parent to all of
subordinate forms, as in this set of definitions:
define tosee-1
label to see
child sense-1
define see-1
label see
child tosee-1
number plural
tense present
person third
define sees-1
label sees
tense present
person third
number single
child tosee-1
define saw-1
label saw
child tosee-1
number single
tense past
person third
There is no requirement
that any of the subordinate forms be present. However, at
a minimum, it is good idea that you define the infinitive
and the third-person-present forms. Notice how the forms
are linked together; the subordinate froms are children
of the inifinitive.
- nouns
- All nouns must be able to
trace their lineage back to things. This means
that, by following child links, you should be
able to find things as an ancestor.
Single and plural forms
of nouns may be linked such that Brainhat understands
their relationship to each other
define spider-1
label spider
number single
child spiders-1
define spiders-1
label spiders
number plural
child insect-1
Above we see how spider
is related to spiders; the former is the child of
the latter.
-
prepositions
- All prepositions must be
able to trace their lineage back to prepositions.
This means that, by following child links, you
should be able to find preposition as an
ancestor.
-
attributes
- All attributes must be
able to trace their lineage back to attribute-1.
This means that, by following child links, you
should be able to find attribute-1 as an
ancestor.
- articles
- All articles must be able
to trace their lineage back to article. This
means that, by following child links, you should
be able to find article as an ancestor.
|
