Metadata, Resources, and the Resource Description Framework 3 The XML aspect is important for several reasons. XML is as close to a global, universal data format as we come today (because it uses Unicode, it is more universal than ASCII, which is restricted to the Latin alphabet). How to handle it in databases, how to transport it over the network, and how to build applica- tions that use it are all well understood. And it has a large installed base, with plenty of applications. RDF builds on XML to create descriptions, and descriptions are metadata: data about data. It can be very hard to understand, and there are basically two ways of explaining it: As object properties, or as profiles. Which is more useful to you depends on your background. First, let us start with the concept of descriptions. A description of a document is a document in its own right. Documents are nothing but a sequence of frag- ments, elements of information, and the order and structure of the fragments constitute a metadescription of the document. Of course, there can be other descriptions as well, such as what the document is about, how it should be pre- sented in different formats, and anything that pertains to the document but is not the document itself. In RDF, you always identify the object you are describing by a unique address, the Universal Resource Identifier. The descriptions can be object identifiers as well (URIs). This means that in object-oriented terms, you can describe which classes an object belongs to and then compare the listing of classes for this particular object to other objects and so find out what they have in common. The descriptions are also data. Data about data is data, too. And metadata is nothing but data about data. It can be embedded in the document, or exist sep- arately from it, as a document or as headers in a protocol, for instance. But there is nothing that limits the use of metadata to documents. It is possible to describe any object using metadata. And anything can be an object, from the collection of all information in the universe down to the letters on this page. When the metadata about the object is structured to provide a description, and the structure is common for all instances of the same type of object, it is a pro- file. The profile can have different values for different instances, but the struc- ture is always the same. So, all the books in the library can be described using the same library cards, but each library card will have different content, even though they all include book title, author, and so forth. What falls under one classification to one man, however, is something else to another. No object falls unambiguously into a single classification. Not even in physics can we find unambiguous ways to describe the objects we talk about. What is a quark? Is an electron really a wave or a particle? And it just keeps going uphill from there. Language,  from  the  point  of  view  of  the  language  philosopher,  is  based  on  a social contract between the sender and receiver. If I am speaking Japanese and 69528_CH01Ix  4/6/2001 8:15 AM  Page 3