10 C H A P T E R   1 schema provides property names and property values, as well as constraints on them that enable their use in calculation. To a computer, a vocabulary is just a set  of  objects  with  specialized  properties:  one  word  can  only  be  a  figure, another can only be five characters long, and so on. A computer cannot have a relation to the vocabulary as such. It is silly that I have to point this out, but the anthropomorphization that follows on the use of intelligent for agents means that users tend to ascribe to them properties that they do not have. Large parts of this book will be devoted to demystifying the relationship between agents and data. The artificial intelligence field is important in many aspects, not in the least because it serves to identify the problem space. There are four problems that, in translation from concept to practice, have dogged intelligent agents: nn   Data encoding nn   Data transport nn   Resource discovery nn   Drawing conclusions based on data The artificial intelligence work has largely concentrated on the last one, and no practical systems exist that solve the first three problems in a distributed fash- ion.  Using  the  technologies  developed  on  the  Web,  however,  all  four  of  the problems become simple. Applying technologies from the World Wide Web to artificial  intelligence  systems  solves  problems  that  have  dogged  researchers for the last twenty-five years, at least. There have been a number of vocabularies that attempted to describe special- ized domains before RDF was invented. Most of them are specialized vocabu- laries defined for an area of artificial intelligence. There are a few examples of other types of vocabularies where an attempt has been made at formalizing a terminology. (There are probably as many terminologies as there are disciplines, and  most  specialized  domains  of  knowledge  have  their  own  terminologies: Engineers  talk  about  nuts  and  bolts  in  a  particular  way,  whereas  a  theater director will have a precise vocabulary to describe the action on the stage.) The examples include the rules for games, which are formalized very precisely to determine which types of clubs are allowed in golf, or how you can touch the ball in soccer. These rule books contain very precise definitions of the different aspects of a game. But they are not concerned with information objects; if we were talking about a sports service, it might make sense to define the vocabu- lary in terms of the rule book for the game, for instance.   RDF gives you a format for describing information objects, but it does not say anything  about  what  terms  you  should  use  to  make  the  statements  that describe the objects, and what those terms mean. It does, however give you a 69528_CH01Ix  4/6/2001 8:15 AM  Page 10