Page 76 -
P. 76
2.3 Conceptual models 45
You asked: How many legs does a ceyipede have?
Jeeves knows these answers:
Where can I find a definition for the math term
leg?
,.
Where can I find a concise encvclo~edia article on ?
centipedes?
-
Where can I see an image of the human
appendix?
Why does my leg or other limb fall asleep?
Where can I find advice on controlling the garden pest ?
millipedes and centipedes?
Figure 2.4 The response from "Ask
ources from Britannica.com on Jeeves for Kids!" search engine when
asked "how many legs does a cen-
tipede have?"
way the child expects. For example, a child might type in a seemingly simple question,
like "How many legs does a centipede have?" which the search engine finds difficult
to answer. Instead, the search engine replies by suggesting a number of possible web-
sites that may be relevant but-as can be seen in Figure 2.4-can be off the mark.
Another problem that can arise from a conversational-based, conceptual
model is that certain kinds of tasks are transformed into cumbersome and one-
sided interactions. This is especially the case for automated phone-based systems
that use auditory menus to advance the conversation. Users have to listen to a
voice providing several options, then make a selection, and repeat through further
layers of menus before accomplishing their goal (e.g., reaching a real human, pay-
ing a bill). Here is the beginning of a dialog between a user who wants to find out
about car insurance and an insurance company's reception system:
<user dials an insurance company>
"Welcome to St. Paul's Insurance Company. Press 1 if new
customer, 2 if you are an existing customer".
<user presses 1>
"Thank you for calling St. Paul's Insurance Company. If you
require house insurance press 1, car insurance press 2,
travel insurance press 3, health insurance press 4, other
press 5"
<user presses 2>
"You have reached the car insurance division. If you re-
quire information about fully comprehensive insurance press
1, 3rd-party insurance press 2 . . ."