Statements
A statement is a statement of fact, also just called a fact. They are the bread and butter of Pyke. Statements are the data values that Pyke acts upon.
You might also think of a statement as a spoken sentence. For example, the Pyke family_relations example deals with sentences like:
"Bruce is the son of Thomas (his father) and Norma (his mother)."
But we condense the sentence down to it's essence. In this case, the sentence revolves around three things: Bruce, Thomas and Norma. All of the rest of the words can be condensed into a single identifier that identifies the relationship between these three things:
son_of(Bruce, Thomas, Norma)
We can give these condensed sentence structures any names that we want. In this case, I chose son_of. I might have also chosen "parents_of", which might conjure the following English sentence:
"The parents of Bruce are Thomas (his father) and Norma (his mother)."
Or:
"Bruce's parents are Thomas (his father) and Norma (his mother)."
But the son_of form carries the additional information that Bruce is a son rather than a daughter. So this is the form used in the family_relations example.
Caution!
Statements are not functions! When we wear our Python hats, son_of(Bruce, Thomas, Norma) looks like a function call! We might expect that it can be executed to do something and possibly return a value. But when we wear our Pyke hats, this is just a statement, or a piece of data. It doesn't do anything and it never returns a value!
Note that it makes perfect sense to have several statements defining the same relationship between their arguments:
son_of(Bruce, Thomas, Norma) son_of(Michael, Bruce, Marilyn) son_of(David, Bruce, Marilyn)
But this only makes sense if they have different arguments. There is never a need to state the same fact twice. Thus we can never establish two facts (two statements) that are identical. If we try to do this, the second one is silently ignored.
So:
son_of(Bruce, Thomas, Norma) son_of(Bruce, Thomas, Norma) son_of(Bruce, Thomas, Norma)
is exactly the same as:
son_of(Bruce, Thomas, Norma)
Finally, we see that the position of each argument is important. In our son_of example, the meaning of each argument is:
son_of(son, father, mother)
Thus, changing the order of the arguments changes the meaning of the statement. So:
son_of(Bruce, Thomas, Norma)
and:
son_of(Bruce, Norma, Thomas)
mean different things! The first statement says that Thomas is the father of Bruce, but the second statement says that Norma is the father!
Syntactic Structure of Statements
So we see that statements in Pyke are very structured.
Pyke categorizes statements into knowledge bases. You create knowledge bases to help you organize your statements. A knowledge base in Pyke roughly corresponds to a module in Python.
Note
Pyke does not allow knowledge bases to contain other knowledge bases, only information about statements. Thus, there is only one level of knowledge bases; and beneath each knowledge base, one level of statements.
So statements have three components:
- The name of a knowledge base. For example, family.
- The name of a knowledge entity. For example, son_of.
- The statement arguments. These are just Python data. Currently in Pyke, there is a push for these arguments to be immutable.
The syntax for a statement looks like this:
statement ::= IDENTIFIER '.' IDENTIFIER '(' {argument,} ')'
Knowledge Base
The first IDENTIFIER is the name of the knowledge base. In our family_relations example, this is family.
Note
You'll see that within backward-chaining rules, the name of the knowledge base may be omitted. It defaults to the currently selected rule base for this rule base category. You'll learn more about this later.
Knowledge Entity
The second IDENTIFIER is the name of the knowledge entity. This is the relationship between the arguments. You could also think of this as the statement type or topic. For example, son_of is a type of statement with three arguments: (son, father, mother). Or the (son, father, mother) arguments are about the topic son_of.
Arguments
The arguments can be any simple Python data value (numbers, strings, None, True or False) or tuples of these values (including nested tuples). Currently, statements are supposed to be immutable, so all of the arguments are immutable. The arguments relate to the topic, above, to make a complete statement.
Note
Prolog allows arguments to be other statements (functors). But Pyke needs to integrate into Python and Python has no concept of a "statement". So we just use tuples in Pyke because Python is very happy with tuples!
So the complete statement for our family_relations example is:
family.son_of(Bruce, Thomas, Norma)