Proceedings

Abstract:

Think of an XML document, and the chances are that you will be thinking of a series of nested tags neatly indented to show the structure of the data being presented. The average XBRL (eXtensible Business Reporting Language ) document, however, is almost entirely flat - a series of data items all at the same level. All the information that would normally be encoded into the hierarchical structure of a standard XML document has been stripped out, and resides in an XBRL "taxonomy" that defines the concepts being used and their relationships to one another. These taxonomies make heavy use of in combination with XML Schema to define fluid and extensible frameworks for reporting different classes of business data.

In this paper I first of all give a brief description of how these mechanisms work. I then go on to examine how and why XBRL takes an approach that initially seems so counter-intuitive, looking at the advantages of disadvantages of its separation of structure from content from a number of angles, particularly validation, information reuse and versioning. While it is undoubtedly true that there are some areas of the specification that are very specific to the primarily financial domain that it addresses, and others that have proved to be unexpectedly complex to implement and work with, I believe that some of the core ideas in XBRL offer some interesting insights into many problems faced by the wider XML community. For example, extending existing content models whilst maintaining backward compatibility is an area that has exercised many minds; through its use of an extensible graph structure rather than a traditional hierarchy, XBRL offers a novel and interesting solution. By drawing out positive lessons like this, and negative ones where XBRL has taken routes that led it into difficulties, this paper aims to be of interest both to those looking to work with XBRL itself, and to those developing and using similar technologies that may be able to take advantage of some of the techniques it is based on.