nbul 2004 
10, Czech 
formation 
| Nigeria, 
litics and 
S (PISTA 
rch 
nation 
sin, 
Class of 
pp. 761- 
upport 
n 
ation for 
ernational 
1391. 
ment and 
"ironment 
0, Brazil, 
id Social 
P 1007 
Iniversity 
1 
4019.ht 
eduction 
d Cross- 
AGENT-BASED PERSONALIZED TOURIST ROUTE ADVICE SYSTEM 
Yuxian Sun?, Lyndon Lee" 
"Department of Geo-Information Processing, International Institute for Geo- 
Information Science and Earth Observation (ITC), P.O.Box 6, 7500 AA Enschede, 
The Netherlands 
(email: sun@itc.nl) 
"Department BT Exact, Adastral Park, Martlesham, IPSWICH, IP5 3RE, 
United Kingdom 
(Iyndon.lee@bt.com) 
KEY WORDS: Automation, data mining, navigation, GIS, mathematics, knowledge base, requirements 
ABSTRACT 
This paper proposes a tourist route model that is based on the integration of GIS spatial analysis functions and a 
kind of heuristic search algorithm based on local optimisation. A vector-based model is used to represent tourists’ 
personal interests and the available tourist resources. Based on these two representations, the system can build a 
user model that indicates the attractions values of the tourist features in a particular area to individual tourists. The 
route agent then uses this user model to generate personalized tourist routes. This research adopts a Tabu search 
method, called extension/collapse algorithm, that is applied in the operations research for maximizing some 
utilities under certain constraints. It is a heuristic search method that progressively selects each tourist site based on 
both its attraction value and the cost value. An empirical evaluation has been carried out on the personalized route 
advice system. The result suggests that the tourist route model generates tourist routes that are empirically 
consistent with the tourists’ preferences. 
INTRODUCTION 
As mobile devices decrease in price and size, and 
increase in power, storage, connectivity and 
positioning capabilities, tourists will increasingly 
use them as electronic personal tour guides. 
However, customers are not satisfied by being 
dumped with a lot of non-relevant information on 
their mobile devices. “Context-aware” and 
“personalisation” have become the buzzwords in 
location-based services. 
The context of a tourist may include external 
factors, such as physical geographical environment 
and social-cultural events, and internal factors, such 
as the physical conditions of a tourist. 
Personalisation allows tourists to express what they 
like and the computer program will use this 
information to plan tour routes and activities for 
tourists. Therefore, knowing what each tourist likes 
and using this information to provide a personalised 
services become the main challenge for the service 
providers. 
One of these services is the personalised tourist 
route recommendation. Most of the GIS network 
analysis tools, such as ArcGIS, and navigation 
Systems are able to recommend routes to travellers. 
However, they mostly apply the shortest-path based 
on Dijkstra's shortest distance algorithm in time or 
distance. These functions cannot be directly applied 
to tourist route calculation problem because of the 
following two reasons: 
319 
e A shortest or a minimum-cost path is not, in 
most of the cases, what a tourist needs. Instead, 
he would like to follow a route that can give 
him the most satisfaction by including as much 
as possible those features that he likes. 
e A tourist usually has certain kinds of resources 
to be consumed during a tour, which can be the 
time or distance that the tourist can afford 
during a tour. This resource is usually 
expressed as an upper bound that cannot be 
exceeded, but should be consumed as much as 
possible during a tour. 
To overcome these problems, this research adopts a 
different algorithm, e.g. a kind of Tabu search 
method, that is applied in the operations research for 
maximizing some utilities under certain constraints. 
It is a heuristic search method that progressively 
selects each tourist site based on both its attraction 
value and the cost value. 
SYSTEM ARCHITECTURE DESIGN 
The design of such a personalised tourist route 
advice system is based on agent-based technology. 
Agent technology has been mostly adopted by the 
software that simulate individuals’ behaviour in a 
social or physical environment (Itami and Gimblett, 
2001; Lee et al., 2002; Wasson et al., 2001; Zipf, 
2002). The individuals can be human, animals, 
plants or other objects that possess certain behaviour 
acp mA epum 
  
EEE DENE TEE TIGE