l~ecommending satisfactory routes for driving requires data about the road network and an individual’s relative weighting of available factors. We describe an interactive planning system that generates routes with the help of a driver and refines its model of the driver’s preferences through interaction. Results of a study indicate that it is possible to model drivers through feedback about relative preferences, but a richer description of the road network can improve accuracy. Our adaptive route advisor unobtrusively collects data on preferences in relevant areas, provides its user with a useful service, and improves its performance as it updates its user model.
[1]
Karen Zita Haigh,et al.
Route Planning by Analogy
,
1995,
ICCBR.
[2]
Pat Langley,et al.
Personalization of the automotive information en-vironment
,
1997
.
[3]
James F. Allen,et al.
TRAINS-95: Towards a Mixed-Initiative Planning Assistant
,
1996,
AIPS.
[4]
Greg Linden,et al.
Interactive Assessment of User Preference Models: The Automated Travel Assistant
,
1997
.
[5]
Kazuo Miyashita,et al.
Case-Based Acquisition of User Preferences for Solution Improvement in Ill-Structured Domains
,
1994,
AAAI.