Knowledge Management and Specialized Information Systems - Principles of Information Systems - page 454

Natural Language Processing and Voice Recognition
As discussed in Chapter 4, natural language processing allows a computer to understand
and react to statements and commands made in a “natural” language, such as English. Google,
for example, has a service called Google Voice Local Search that allows you to dial a toll-free
number and search for local businesses using voice commands and statements. 46 Restoration
Hardware ( www.restorationhardware.com ) has developed a Web site that uses natural lan-
guage processing to allow its customers to quickly find what they want. The natural language
processing system corrects spelling mistakes, converts abbreviations into words and com-
mands, and allows people to ask questions in English.

natural language processing
Processing that allows the
computer to understand and react
to statements and commands made
in a “natural” language, such as
English.

Dragon Systems' Naturally
Speaking 9 Essentials uses
continuous voice recognition, or
natural speech, allowing the user to
speak to the computer at a normal
pace without pausing between
words. The spoken words are
transcribed immediately onto the
computer screen.
(Source: Courtesy of Nuance
Communications, Inc.)

In some cases, voice recognition is used with natural language processing. Voice recogni-
tion involves converting sound waves into words. 47 After converting sounds into words,
natural language processing systems react to the words or commands by performing a variety
of tasks. Brokerage services are a perfect fit for voice-recognition and natural language pro-
cessing technology to replace the existing “press 1 to buy or sell a stock” touchpad telephone
menu system. People buying and selling stock use a vocabulary too varied for easy access
through menus and touchpads, but still small enough for software to process in real time.
Several brokerages—including Charles Schwab & Company, Fidelity Investments, DLJdi-
rect, and TD Waterhouse Group—offer these services. These systems use voice recognition
and natural language processing to let customers access retirement accounts, check balances,
and find stock quotes. Eventually, the technology will allow people to make transactions
using voice commands over the phone and to use search engines to have their questions
answered through the brokerage firm's call center. Using voice recognition to convert record-
ings into text is also possible. 48 Some companies claim that voice-recognition and natural
language processing software is so good that customers forget they are talking to a computer
and start discussing the weather or sports scores.

Learning Systems
Another part of AI deals with learning systems , a combination of software and hardware that
allows a computer to change how it functions or reacts to situations based on feedback it
receives. 49 For example, some computerized games have learning abilities. If the computer
does not win a game, it remembers not to make the same moves under the same conditions
again. DARPA is investing about $10 million into a learning system called Bootstrapped
Learning that will help military computers learn from human instructors. 50 If successful, the
project could help the development and control of unmanned aircraft. The Center for Au-
tomated Learning and Discovery at Carnegie Mellon University ( www.cald.cs.cmu.edu ) is
experimenting with two learning software packages that help each other learn. The hope is
that two learning software packages that cooperate are better than separate learning packages.
Learning systems software requires feedback on the results of actions or decisions. At a
minimum, the feedback needs to indicate whether the results are desirable (winning a game)
or undesirable (losing a game). The feedback is then used to alter what the system will do in
the future.

learning systems
A combination of software and
hardware that allows the computer
to change how it functions or reacts
to situations based on feedback it
receives.