DTD/Schema is a set of computer readable languages that can be used to describe the structure and format of a markup language. It is used today to define the structure of all markup languages.

The idea is that you can define your own markup language specific to your needs.

For example, let say I was trying to write an application that stores and displays addresses.

Therefore a sample of the data I am storing would be:

123 Main St. Apt 210. Somewhere IA 12345

To a computer this line above really means nothing besides that it is a string of text.

Now if I were to use DTD and a DOCTYPE, I could store the information like this:

<address>
  <street>123 Main St.</street>
  <apartment>210</apartment>
  <city>Somewhere</city>
  <state>IA</state>
  <zipcode>12345</zipcode>
</address>

In this format I bring more meaning to the data by wrapping it in a custom markup language. The data becomes much easier for a computer to extract specific parts of it. Also as a human reading over the data, the markup brings more human readable meaning.

We have effectively bridged a gap between the human and the computer. We have brought more meaning to a human reading the raw data and we have enabled the computer to analyze and break down the data in the same way a human would.

DTD/Schema and DOCTYPE would be used to tell the computer that inside of the tag ‘address’ there are tags ‘street’, ‘apartment’, etc.

My comments about the application of this in the arena of artificial intelligence were simply remarking on the theory this provides. The fact that we have a computer language that can define the syntax of another language. I think that this idea could someday grow into use with human spoken language.