Architecture for Including Personalization in a Mobile GIS via Semantic Web Techniques - Geographical Information Systems: Trends and Technologies - page 231

the best data for each user considering different preferences, taking also
into account context, i.e., we propose a CARS. Thanks to this algorithm,
we give different and personalized routes to each user.
In the following we present an overview of the algorithm. This
algorithm uses mainly knowledge-base fi ltering, but corrected by content-
based fi ltering. It is based on two key points:
1) It only offers viable routes, namely, the algorithm suggests a visit only
when the user can carry it out. For example, it does not offer to visit a
place that would be closed at the expected arrival time.
2) The proposed route does not necessarily suggest the maximum number
of attractions to visit, because the visitor may want to visit fewer POIs,
but mainly those that are more interesting to him/her. Thus, from
a collection of viable routes, the algorithm will choose those most
attractive to every single user.
Considering these two features, we designed an iterative algorithm.
For each step we indicate, in parentheses, from which ontology the data
is obtained:
1. From a group of POIs inside a region, a set of random routes
doing combinations of those POIs is generated ( tourism ontology
—POI information, personal ontology —temporal and geographic
preferences).
2. POIs are ordered by distance ( tourism ontology —POI location)
3. Routes that are in one of the following cases, are discarded:
• Case 1: the needed time to fi nish the route exceeds the available time
for the visitor ( historical of routes ontology—POI average length of
visit)
• Case 2: there is a POI closed when the user arrives to visit it ( time
ontology —POI timetable)
• Case 3: there is no place to eat when the visitor wants to eat—the
visitor decides which meals he/she wants to do ( time ontology —POI
timetable, personal ontology —content preferences)
4. Once there is a group of viable routes, the algorithm selects the most
appropriate to the user needs considering the following parameters:
a. POI average rating: given by other visitors ( tourism and historical of
routes ontologies)
b. Companions of the user: according to who is travelling with the user
different kind of POIs will be suggested. For instance, a night club
has a higher rate when travelling with friends ( personal , tourism and
historical of routes ontologies)
c. Kind of POIs preferred by the user ( personal and tourism
ontologies)