7/31/2019 bi2 - Copy

1/31

ALBERTO SALGUERO, FRANCISCO

ARAQUE, CECILIA DELGADODepartment of Software Engineering ::

ETSIIT

University of Granada (Andaluca)

C/ Periodista Daniel Saucedo Aranda s/n

SPAIN

agsh, faraque, cdelgado {@ugr.es}

Application of

Business Intelligence

methods for

personalizing tourist

services

7/31/2019 bi2 - Copy

2/31

Outline

7/31/2019 bi2 - Copy

3/31

Introduction

Planning trip is complicated

and need some expertise.

Grown of internet provides

are a lot of information that

should be read prior of

selecting the most

interesting spots

7/31/2019 bi2 - Copy

4/31

Contd

tourism in Spain has

considerably contributed to its

economy, representing the

108 % (106.374 millions of

Euros) of the GDP of the

country.

The Spanish government,

concerned with this issue has

incentivized the development

of projects which aims to raise

the quality of the touristic

services.51.6% people seek

information about their trip

through internet

7/31/2019 bi2 - Copy

5/31

Contd

Spanish government has impulse the projects relying onthis technology, giving as result, for instance, the

touristic information systems

However, there is another important issues

web sites should incorporate some kind of decision support

functionalities to assist users in selecting the most appropriate

content

7/31/2019 bi2 - Copy

6/31

Contd

Common tool for planning trip:

Google Earth/map, GPS navigators, Microsoft

MapPoint/Virtual Earth and other mapping

tools.

BUT The problem of using these tools is

that they always give the shortest path

between locations. This is useful in mostof cases but it is not when planning a trip!!.

7/31/2019 bi2 - Copy

7/31

Contd

Need system for solving this following

issues:

make easy the selection of the most

interesting spots and

planning the trip according to the

characteristic of the customer.

DATA MINING TECHNIQUES

7/31/2019 bi2 - Copy

8/31

Contd

The system flow is based on followingdata mining techniques

7/31/2019 bi2 - Copy

9/31

Contd

This system uses a clustering process for finding theprofiles of the customers and then we use this

information for offering him the most interesting points of

interest according to the historical data recorded by the

system.

Once the customer has selected the POIs he wants to

visit the system is able to find the route between them

which better fits to his personal characteristics (age,

overall physical condition) and to other environmentconditions like weather forecast or timetables.

The assumption is that similar customers areinterested in the same things.

7/31/2019 bi2 - Copy

10/31

Data Warehouses and Data

Mining

Data is vital asset for any organization

Must be available anytime

data by themselves are useless, they mustbe put together to produce useful

information. In turn, information becomes

the basis for relational decision making.

Need DATA WAREHOUSE

7/31/2019 bi2 - Copy

11/31

Contd

A DW is a database that stores a copy ofoperational data which structure is

optimized for query and analysis.

DW Scope> Entire enterprise Reduced Scope -> Datamart> Single

Department

7/31/2019 bi2 - Copy

12/31

Contd

The generic architecture of a DW is illustrated below

The data are extracted from the sources and then loaded

into the DW using various data loaders and ETL tools

The warehouse is then used to populate the varioussubject (or process) oriented data marts and OLAP

servers.

7/31/2019 bi2 - Copy

13/31

Contd

Data Mining processes

assists to find out the

patterns, features and in

general the knowledge we

are looking for. this projectused a process of clustering

and classification.

for finding homogeneous

groups of individuals andacting according to the

knowledge about the group.

7/31/2019 bi2 - Copy

14/31

System architecture

DW system + OLAP functionalities provide solutions formarketing problems, since it transforms operational data

into strategic decision-making information.

Following is the architecture of the system

7/31/2019 bi2 - Copy

15/31

Contd

main functionality of each module is:

Extract, transform and load: This module isresponsible of extracting and loading the

customer personal information and theinformation the system needs to per-form the

planning of the trip (public transport timetable,

weather forecast) in the DW.

7/31/2019 bi2 - Copy

16/31

Contd

Clustering: Given a list of users, this module iscapable of performing a clustering process, obtaining

the list of customer profiles as the result. The

customer profile includes several variables. Although

they are advised to, the customers can freely fill or noteach variable value.

Age

Salary

Marital

Number of Family Members

Study Level Achieved

Overal Physical Condition

7/31/2019 bi2 - Copy

17/31

Contd

Classifier: Once a customer has beenregistered he is ready for planning the trip

but, firstly, he has to selects the spots he

wants to visit. One of the objectives of thesystem is to suggest to the customer the

spots he might be interested in, avoiding

the bothersome task of browsing the entirespot data base.

7/31/2019 bi2 - Copy

18/31

Contd

Matcher: Once the system has the profileof the user it has to offer him the spots he

might be interested in.

Route adapter generator: this module isresponsible of creating, once gathered the

spots the customer wants to visit, the route

for the trip.

7/31/2019 bi2 - Copy

19/31

Planning the trip

focus on the process of generating theroute which best fits to the customer.

Once the customer has selected the spots

he wants to visit, the next step is to find apath connecting them.

The shortest path is not always the best

option -> For this reason, one of thefundamental parts of the system is the

process of collecting information about the

area to visit.

7/31/2019 bi2 - Copy

20/31

Contd

There can be more than one direct path from a spot to

another one.

Record all information about how to go from every spot

to the rest. Use matrix containing these direct paths.

The users of the system play an active role in this part of

the system because they are encouraged to add more

direct paths from the set of spots as well as defining new

spots.

7/31/2019 bi2 - Copy

21/31

Contd When defining new paths between spots it is necessary

to specify their metadata. This information is available tothe rest of the users and can be used for generating

more personalized routes.

Actually, every element of the matrix is a vector

containing a value for each of the metadata properties: Distance

Time

Transport

Mean unevenness, Maximum unevenness Roughness

Shopping

Shade

7/31/2019 bi2 - Copy

22/31

Contd

This system adopts pseudo random based solution. =>

time cost effective and able to deal with multiple user

concurrently.

For the evaluation process, the user has to specify the

importance every link property has for him. By default,

these values are set to the mean values given by the

users in the same cluster the customer belongs to.

7/31/2019 bi2 - Copy

23/31

Granatum project

Web based application

A prototype developed by AndalusiaResearch Program

Aims to enhance tourism quality service inAndalusia, Spain

Recording registered user characteristic, sothe user can choose: see the predefinedroutes, create a customized route or generatean automatic personalized and adaptedroute.

7/31/2019 bi2 - Copy

24/31

Option : Predefined Routes

predefined routes is defined by thesystems administrator

These routes are static set of POIs and

links between those POIs. The routes presented in this option

represent the typical routes any touristic

guide offer to the tourist. Provide general idea of the location he is

going to visit.

7/31/2019 bi2 - Copy

25/31

Option: Customized Routes

Designed for tourist who wants to

personalized their tourism routes.

Consists of following steps:

Selecting the Points Of Interest

Determining the user preferences

Generating the personalized routes

7/31/2019 bi2 - Copy

26/31

Selecting the Points Of Interest

With this option, user may choose POI that he wants to

visit.

Assists the user to find interesting spot for tourist easily.

7/31/2019 bi2 - Copy

27/31

Determining the user

preferences Once POI has been chosen, next step is generating routes

User is given list of combo box with preference value (indifferent,

little important, important, very important, essential)

Depending on the kind of information the parameter represents it

can have a direct or inverse relation with respect to the score of theroute.

7/31/2019 bi2 - Copy

28/31

Generating the personalized

routes Once the system has the POIs and the preferences of

the user a list of routes is generated

The personalized routes are sorted according to their

score with respect to the preference values given by the

user.

Option : Automatic personalized

7/31/2019 bi2 - Copy

29/31

Option : Automatic personalizedand adapted routes

No need express the POIs wish list and preferences

In this case a number of POIs the most included by similar tourist

in their routes are selected by default and the routed adaption to

the tourist preferences are carried out

similar tourist assumed has similar characteristics.

7/31/2019 bi2 - Copy

30/31

Viewing the routes

Detail can be obtained in form of text, map

or some kinds of mapping application files

format (Google Earth or GPS navigation

systems like TomTom.)

7/31/2019 bi2 - Copy

31/31

Kesimpulan

DW dalam kasus ini dapat digunakan untukmembantu user menemukan tempat-tempat yangmenarik bagi user untuk dikunjungi besertarutenya yang sesuai dengan karakteristik user

Teknik Data Mining yang digunakan di bidangTourism, pada faktanya dengan pendekatanclustering dan clasification yang dipasangakandengan DW menunjukan teknik yang powerful

untuk mengantisipasi permintaan user danmemberikan informasi yang mereka butuhkan.