The Web was initially a Web of documents, documents
designed to be read by people. The companies that provided
the document management infrastructure for the Web, however,
were not the companies that had dominated traditional pre-web
document management (IBM/Lotus, Documentum, Filenet).
They were a new breed of company focused explicitly
on web-based access to documents — Inktomi and
Google.
Today corporations are moving their IT infrastructure onto
the Web. They are constructing SOA-based applications
that talk to one another across the Web. These applications
are not based on documents. They work with data. And
there are no people in the loop to interpret the data.
SOA applications need a standard for sending information
from one computer to another in a way that allows its
meaning to be understood by the receiving computer system.
Low-level, record-oriented standards like
SQL do not stand up to this requirement. Too much of
the meaning of the data is only implicit in the column
headings, or worse, buried in the applications that use
the data.
During the mid-1990's the emergence of the Web drove standards for specifying the format of documents that were exchanged between people across the Web: HTML and XML. Now the Web is driving a new set of standards for specifying the meaning of data to be exchanged between computers across the Web: RDF and OWL.
The XML standard reached full W3C recommendation status in 1998. It has since become ubiquitous.
The OWL standard reached full W3C recommendation status in 2004, six years later. We think that it will become the base for a new generation of web-centric data management software over the 2005-2015 decade, paralleling XML's gradual take-over of web-centric document management in the 2000-2010 decade.
Just as document management for the Web came from new companies focused on how the qualifier ‘for the Web’ changed everything, so data management for the Web may come from new companies focused on how that qualifier changes everything for data management.
The query language for data in the Web will not be SQL. It may be a Google search bar — but that bar will be driven by an ontology-based understanding of what the question means, rather than by a set of key words.
If the question is, “Who will be to Oracle, what Google was to Documentum?”, we want the answer to be “DATA-GRID”.