Digital Signal Processing Reference
In-Depth Information
Figure 3-4.
Sightseeing areas employed in the task (Phase I and II).
3.2.2
The Task and Instructions
A preliminary survey of 302 drivers indicated that the three most desirable
types of driving information are traffic information, restaurant information,
and sightseeing information. As a result, data was gathered for three
predefined dialog tasks:
Task 1: Obtain route guidance to a particular destination.
Task 2: Find a restaurant for lunch.
Task 3: Find tourist attractions to visit after lunch.
The speaker was instructed to pretend that he/she was in a car equipped
with an in-vehicle terminal, traveling on an overnight trip to one of two
sightseeing areas outside Tokyo. The number of possible destinations was
limited to two because it was difficult for the operator to prepare sufficient
driving information for multiple sightseeing destinations in advance. The
chosen destinations were chosen to be sufficiently popular to motivate the
speakers to generate ample questions. In addition, the destinations were
chosen to be not too far from Tokyo (destinations that are far from Tokyo
require use of expressways, which renders route guidance less useful).
Keeping these constraints in mind, we selected Hakone and Izu as the
possible tourist destinations (see Figure 3-4). Both of these sightseeing areas
have many hot springs and tourist attractions, and are very popular to
residents in the Tokyo Metropolitan Area. There are several ways to reach
each of these areas from Tokyo.
