WebReference.com - Excerpts from chapter 7 of The Intelligent Wireless Web, from Addison-Wesley (1/4)
The Intelligent Wireless Web
MERGING AI WITH THE WEB
Intelligence usually refers to the ability to reason, to solve problems, to remember information, or to learn new ideas. So, as we begin to add more intelligent agents, smart applications and Artificial Intelligence (AI) programs to Web sites, could we ask, "Does the Web have an Intelligence Quotient (I.Q.)?"
The simple answer is, "No, at least not explicitly." The IQ measure for humans was developed based on the rates at which intelligence develops in children. Its original definition is that IQ is a measure of intellectual development determined as the ratio of a child's mental age to his chronological age, multiplied by 100. The scale is extended to adults where an IQ of 100 relates to the performance of an average adult on a suitable standardized-test. For humans, IQ is a good predictor of various measures of success or failure in life, but making computers that can score high on IQ tests would be only weakly correlated with their usefulness. Consider the ability of a child to repeat back a long sequence of digits. This correlates well with other intellectual abilities of the child, because it measures how much information the child can remember and use for computation. However, repeating back a lengthy information string is a trivial task for even extremely limited computers. Nevertheless, some of the problems on IQ tests present useful challenges for Artificial Intelligence (AI) programs.
So, how will adding intelligent agents, smart applications, and AI programs to Web sites, contribute to the development of the Intelligent Wireless Web? Will they support the development of a network that learns?
To begin to address these questions, we will have to explore some uncharted territory and face some probing and provocative questions, such as:
- How smart are Web applications today?
- What is Web intelligence?
- How does the Web learn?
- Where does Web intelligence reside?
In the following sections, we will begin the discussion of how to introduce intelligence to enlighten the optical pathways that inhabit the Web
How Smart are Web Applications Today?
For the most part, the Web can be considered to be a massive information system with interconnected databases and remote applications providing various services. While these services are becoming more and more user oriented, the concept of smart applications on the Web is still in its infancy.
One of the most sophisticated applications on the Web today is the Enterprise Information Portal (EIP) operating with the state-of-the-art Markup Languages to search, retrieve and repackage data. The EIP is in the process developing into an even more powerful center based on component-based applications called Web Services. In this section, we will describe EIPs and the XML standards, frameworks and schema that make up today's most sophisticated and intelligent Web tools and applications. We will also present the tools and competitors developing Web Services. We will consider just how smart Web Services are likely to be. As an example see the inset OpenCYC:
Cyc, named after enCYClopedia was developed over 17 years at Stanford University under the leadership of Doug Lenat. Cyc is a knowledge base containing assertions and inferences that responds to natural language. Developed with $50 millions and over 500 man-years of effort through the Defense Dept. and others, the system today encompasses 1.4 million assertions.
As an experimental project, Cyc emerged as a response to ongoing controversy in the Artificial Intelligence field. Lenat understood the problem that expert systems were "not servants, but idiot savants." What was missing was "common sense."
Within the contentious AI community Cyc is drawing criticism. Cyc is seen to focus on sorting facts and observations into logical categories and even its ability to makes inferences is considered too restrictive.
The release of OpenCyc onto the Web will allow access for free to a limited portion of the Cyc knowledge base. OpenCyc incorporates applications, such as, speech recognition software, database searches, and natural language queries. The full knowledge base (about 20 times the size of the public portion) will be licensed to commercial users. (see www.opencyc.org)
Created: March 20, 2002
Revised: March 20, 2002