Knowledge Representation

Some conjectures about the representation of knowledge are listed below. It is the belief of this author that neither proof nor disproof of these conjectures exist in the literature today. If you know of any proof or disproof of these conjectures please email seth@robustai.net . Applications and implications of these conjectures will not be dealt with in this paper.

Conjectures:

Knowledge can be represented in natural language.
Knowledge can be represented by labeled digraphs.

Please follow this link for a definition a labeled digraphs.

Algorithms can be constructed to compute and generate labeled digraphs from any natural language statements .
Algorithms can be constructed to generate logical inferences of knowledge represented by labeled digraphs and represent that knowledge in labeled digraphs..
Algorithms can be constructed to generate natural language statements of any knowledge that has been represented by labeled digraphs.
Labeled digraphs can be efficiently represented in a digital computer.

Examples of knowledge represented by labeled digraphs:

An example of this is the English sentences: "An apple is a fruit" "Fruit is sweet.".

A more detailed example has been "provided by a six-year-old grade 1 student to describe his knowledge of whales".

Follow this hyper link for an example of the classic problem of Oedipus and his wife is Jocasta answering the question "Does Oedipus know his wife is his mother?" correctly.

The sentence: "The individual referred to by employee id 85740 is named Ora Lassila and has the email address lassila@w3.org. The resource http://www.w3.org/Home/Lassila was created by this individual." can be represented with the diagram under this hyper link from the RDF data model.

Representing labeled digraphs in a computer:

There are many ways to represent labeled directed graphs efficiently in a computer.

One way is with relational databases where each directed vector of the graph is a row in a table. The row would then contain the following attributes: the identity of the source node, the identity of the label node, and the identity of the object node. So the relational database below represents the same thing as the digraph #1 above. I have choose to rename the objects Node, Arrow, Destination node with the attributes Subject Verb Object respectively, and name this knowledge representation scheme SVO.

subject	verb	object
apple	isa	fruit
fruit	is	sweet

Human:

AI:

Another way to represent labeled digraphs in a computer is with the Resource Description Framework (RDF) as recommended by the W3C. The sentence and the digraph in example #4 above can be placed in meta data on the Web with the following markup:

Concluding remarks:

Please note that paper is not proposing using graphical diagrams to represent knowledge inside of computers. Rather it is proposing that graphical diagrams can be used by humans to picture in our minds the underlying deep semantic relationships that are inherent in language and knowledge. Then without loss of precision, these deep semantic relationships can be coded inside a computer where artificial agents can have access to them.

Acknowledgments

Rob Kremer of The University of Calgary for thoughtfully posting his Ph.D. Dissertation Constraint Graphs: A Concept Map Meta-Language on the WWW for me to find with ZurfRider so that I could hyper link this presentation with the already existing "web of science" .
David Longley who's dogged insistence on comp.ai.philosophy that AI was just the "web of science" which was just relational databases which lead me to recognize this connection.