| Cram101 membership is available for a monthly fee.
www.fsunews.com/vnews/display.v/ART/2004/
09/23/4151f0e8c872a
. -
Call Center Computers
A
new AI tool aims to take the frustration out of call centre assistance. Once a
person finally gets through to a customer service representative, often they will
have to wait additional time while the operator searches through the computer
to find the answer they are looking for. IBM is developing a combination speech
recognition utility/search engine that listens in to the support phone calls and
begins searching for the required information before the caller has finished their
request. The tool works by using speech recognition to pick out key words indicating
the caller's trouble. Those keywords are then entered into the call centre database,
giving the operator a head start on pulling up the information.
In
addition to assisting the operator in finding the correct information, the system
can also alert the operator to important points that must be stressed - especially
if those points constitute legal warnings. Just as the system listens to the caller's
end of the conversation, it can also listen to the operator and provide on-screen
reminders in order to ensure that the operator handles the call properly. While
the software is still in its infancy with only a few phrases and words identified,
commercial versions are in the works with a trial version scheduled to be implemented
in a Dutch bank.
www.newscientist.com/news/news.jsp?id=ns99996430
. 
| |
Model of a Multi-Agent System
In
a joint research project from Los Alamos National Laboratory, the University of
Houston and Rensselaer Polytechnic Institute, a model of a network of agents has
been created. The goal is to use the model to predict the behavior of multi-agent
systems - a task that can be very difficult. The model is based on the "minority
game." In this game, agents make decisions and try to be in the minority
when the results are revealed. The agents learn from experience what strategies
are successful. In addition, the researchers created social networks amongst the
agents so that information about strategy could be shared. Therefore, when making
decisions a single agent will consider its own experience as well as the experiences
of its neighbors. The agents strengthen connections with agents that provided
useful recommendations in the past. This can be compared to human social interaction
- we come to value the advice of certain people and thus turn to them preferentially
in the future for support. As a result of these connections, a leadership structure
emerges. To optimize the model, the researchers are adjusting the connectivity
of this leadership so that the influence of those agents is regulated. The information
gleaned from this model could lead to the development of multi-agent systems suited
to applications where human intervention has been required in the past. Since
the agents learn from each other, the human component becomes less important.
www.trnmag.com/Stories/2004/092204/Agent_model
_yields_leadership_092204.html
Network of Robotic Telescopes
With
many astronomical events occurring in a very short span of time, it is important
to be able to react quickly in order to observe them. Anyone who has watched a
meteor shower knows that this is true. If you are not looking in the right place
at the right time you will miss the show. However, if there are many people watching
the meteor shower from many different positions then it is possible to observe
more of the action.
British astronomers are
hoping to take advantage of the information-capturing power of robotic telescopes
by linking them together and controlling them with artificially intelligent software.
The network is called RoboNet-1.0. Since the robotic telescopes in the network
are located around the world, the connection provides a wider range of view as
well as the ability to react quickly to interesting phenomena. Researchers |
