- 15 -
the Web today results in several hundred thousand matches because search engines
don’t have a way of differentiating content due to the limited number of HTML tags (see
Figure 1.3). Instead, they perform searches based on the entire document’s text or, at
best, on the information contained in the
meta tags
. Meta tags are reserved tags in HTML
that are not used for display per se; they are a kind of placeholder for people to use as
they see fit. Many search engines look for information contained in the meta tags when
responding to a query.
Figure 1.3:
Results from a typical search engine.
When XML is widely adopted, search engines will be able to search based on the
content of particular tags, for example
. Let’s say you’re interested in
articles with the
tag of Operations Research, a branch of industrial
engineering. The search results will include only matches to pages that have a
tag and whose contents of that tag include “Operations Research.” In
doing so, you eliminate all the hundreds of thousands of pages that may have the
words o
perations
and
research
mentioned somewhere in the document (for example,
in an article on medicine or defense that’s not relevant for your purpose).
Web-based control systems.
This is one of the most exciting areas for XML
applications. Because your information is now meaningfully structured, you can encode it
to mean anything. Why not use the Web to control your kitchen appliances, bathroom
faucet, and home lighting while you are at the office? Thousands of people will be
creating applications like these in the near future, and our bet is that the background
technology will be XML.
Agents.
This is another of our favorite applications and probably one of the most
fascinating applications of artificial intelligence. Agents are expressions of consciousness
dissociated from your corporeal form. They are essentially software-based assistants that
perform tasks for you based on your individual preferences, for example, performing
specific research for you, controlling your schedule, and managing your stock portfolio.
Agents are still relatively new, and people are doing a lot of great work in this area to use
software to translate human preferences, at the same time conforming to existing
information site structures. Translating human preferences is, generally speaking, very
complicated, and you need a very rich way of describing information before you can
communicate your desires to a piece of software. XML gives developers the flexibility
they need to create more powerful agents.
