Ontology assisted decision making a case study in trip planning for tourism

1
Ontology assisted decision makinga case study in
trip planning for tourism

• Eleni Tomai, Maria Spanaki, Poulicos Prastacos
  and Marinos Kavouras
• National Technical University of Athens,
  Cartography Laboratory, FORTH, Institute of
  Applied and Computational Mathematics

2
Outline

• Objective
• System Architecture
• Ontologies
• Context Matching Algorithm
• General Discussion

3
Objective

• Trip planning in the context of web services.
• Tourists are not a group with homogeneous
  characteristics, therefore personalization is
  crucial in the design of a decision support web
  service.
• Develop a web service that can answer to the
  following types of questions/requests
• I have two days to spend in X, what do you
  propose me to do?
• Today I want to do some sightseeing in X and then
  go to sea.

4
System Architecture - I
5
System Architecture - II

• It consists of
• The two ontologies, namely user profile and
  tourism ontology, implemented using OWL Plug-in
  of Protégé.
• The web-based user interface, which poses
  ontology-driven queries to elicit information
  concerning the user.
• The context matching algorithm which matches the
  user profile to the tourism ontology.
• The map server which shows the location of the
  tourism ontologys concepts of interest, it is
  also utilized to visualize the answer of the
  system.
• The system works in two steps
• the user fills in the interface to generate a
  profile, and then states the question.
• the system runs the context matching algorithm
  and returns the answer.

6
The Ontologies User Profile Ontology I

• The design, concerns agreeing upon the main
  concepts of the ontology along with their
  properties.
• concepts that characterize/ describe a tourist
  (e.g. kind of trip, time available, temporal
  period of the visit, accompanying persons, money
  to spend, transportation means)
• concepts that account for the personal
  information of the user (such as age, gender,
  profession, leisure activities, interests).
• The concepts were further detailed by adding
  sub-concepts for instance, for the concept
  leisure activities the sub-concepts eating out,
  nightlife, shopping and sports were created.

7
(No Transcript)
8
The Ontologies User Profile Ontology II

• Based on each concept, a corresponding property
  was created.
• For instance, from the concept interests, the
  property is interested in can be created assigned
  to the user.
• The properties of the ontology play the role of
  posing questions to the user as a means to elicit
  to information from him/her.
• For the previous example a question is what are
  you interested in? And a possible answer from the
  user is I am interested in museums.

9
(No Transcript)
10
The Ontologies User Profile Ontology III

• The second step is to populate the ontology with
  instances for the concept user.
• Provide an interface to the user so that he/she
  can introduce personal information, interests and
  facts about the visit. The interface resembles
  a questionnaire and, it is web-based.
• The procedure of collecting and recording the
  actual user profiles, is very much guided by the
  predefined user profile ontology.
• When the user is asked to fill in his/her
  interests, he/she can only chose from a list of
  alternatives, given in the questionnaire, that
  correspond to the sub-concepts of the interests
  concept in the generic user profile ontology.

11
(No Transcript)
12
The Ontologies User Profile Ontology IV

• The qualities of this methodology are two fold
• it elicits information on the user profile using
  the same terminology as the one of the generic
  user profile ontology, and
• because the interface is structured based on the
  generic ontology any information introduced
  therein can easily be recorded into the ontology
  as its instance.
• As more tourists use the system, the more the
  ontology gets populated.
• Possible drawback if the concepts of the generic
  ontology are modified, certain elements/pages of
  the interface should change to match the
  ontology.

13
(No Transcript)
14
The Ontologies Ontology of Tourism I

• It includes concepts such as sightseeing,
  shopping, leisure activities etc.
• The concepts of location and time needed have
  central role in the ontology.
• Location refers to either the location a point of
  interest has on a map, or its address, if that
  kind of information is available.
• Time refers to the time it takes for the tourist
  to get to the point of interest plus the average
  time to see the place and come back.
• Other concepts in the ontology concern additional
  information such as accessibility, entrance fees,
  opening hours and the like.
• (Has been implemented for the region of
  Heraklion, Crete.)

15
(No Transcript)
16
The Ontologies Ontology of Tourism II

• Assigning properties to the concepts.
• For instance, from the concept accessibility we
  create the property are accessed by which
  involves the sub-concept of transportation, and
  the concepts which are assigned with this
  property are archaeological sites, museums,
  natural beauty areas etc.
• Properties help us set statements such as the
  following archaeological sites are accessed by
  busses,
• or beaches are accessed by taxis and ferries etc.
  
• The tourism ontology is hidden from the user and
  it is populated with actual data as instances of
  the concepts included therein.

17
(No Transcript)
18
Context Matching Algorithm I

• How it works
• The user queries the system according to his/her
  interests and the time to spend, the semantic
  matching process starts by filtering out the
  services that do not match the service types
  asked by the user.
• The second step involves finding the
  correspondences between concepts and properties
  in the user profile and those in the tourism
  ontology.
• On the third step, additional information
  provided by the user such as visiting period is
  taken into account and narrows even more the
  initial query.
• As for the location of the user, the system works
  under the assumption that they are in the city
  centre of Heraklion.

19
Context Matching Algorithm II

• Provides to the users a list of alternative
  answers and let them make the final decision on
  how to spend their time.
• Crucial feature of the specific algorithm is the
  calculation of time.
• In the tourism ontology the concept of time (time
  needed -tn) encodes the time it takes for the
  tourist to get to a point of interest from the
  centre of Heraklion plus the average time to see
  the place and come back to the centre.
• In the user profile ontology time (time available
  -ta) reflects the available time the tourist has
  to spend in Heraklion.
• The concept of proximity should also be
  incorporated in the algorithm.
• Another concept which should be taken into
  account is transportation means.

20
General Discussion and Future Trends

• Several approaches have been proposed with the
  intension of helping tourisms in exploring points
  of interest.
• Our approach presents novelties such as the
  following
• The system is not a mobile service but a web
  service provided by a local authority.
• The information concerning tourist activities
  (data) is organized in an ontology not separate
  databases.
• The interface where the user inputs his/her
  personal information so that the ontology of the
  user profile is elicited and gradually populated.
• The terminology used in the interface is
  conformant to the terminology of the user profile
  ontology.
• Further work concerns the inclusion of more
  parameters into the context matching algorithm.
• Moreover, several issues concerning the accuracy
  and availability of data should be investigated.