The Lightweight Directory Access Protocol (LDAP) architecture that is currently deployed in numerous servers is built around the older X.500 protocol as a way of using a directory-like notation of retrieve information about a system. Servers that convert LDAP result into XML (or that accept XML X path syntax for retrieving LDAP information) are likely to become commonplace by the time this book is published as the two formats are not dissimilar. Similarly XML can be used to create one or more topic maps into a document space. A topic map can be thought of as a hierarchical table of contents for a whole Web Site. The advantage of a topic map is what it can arrange document by their content rather than their location in a directory.

While the structure here is regular (making it more of a data-centric XML document) its primary purpose is to create a relational map to documents (or conceivably to subdocuments depending on the documents and the server) in what amounts to a concept space. In its simplest form such a space would probably be expressed as a tree-view of some sort but more importantly it is way of abstracting relationship without trying them into an explicit file directory structure. It also reduces redundancy them into an explicit file directory structure. It also reduces redundancy of documents since you don’t need such a document in more than one place to show a relationship. Topic maps and XML-enabled directories will change the way that we look at document management and will ultimately even change the way we think about documents themselves. Since more than one topic map can be assigned to a document space topic map essentially enable the capability of shifting your perspective about the information you have available to you. As we move in caressingly into a data-rich environment (even, perhaps into one where data is considered pollution) XML’s role in document management will become critical. XML as a Data Format

A funny thing happened on the road to XML adoption as the next HTML: several people simultaneously discovered that a table is a data structure. For example consider a simple table of line items from a purchase order. The HTML tells you very little-the structure is a table, the table has headers and rows there are four elements in each row. With a little work you could probably get code to associate each header name with its given column this is meaning that you impose upon the table by all rights the columns could in fact have no association with the column headers in this organization. Suppose however that you could create your own HTML to deal with this problem substituting purchase order for table line item and then labeling each element with the name of the property it contains. Comparing the HTML with POML (for Purchase order markup language) the difference that such a name change makes is dramatic. Just at a glance you can tell that a purchase order is comprised of line items which in turn include names codes unit costs and number of units.