A New Intelligent Tour System Based on Web Mining Technique

ABSTRACT - Now, in Taiwan, people have two days off every weekend. Much more time can be scheduled for leisure activities. Under such circumstances, it seems necessary for everyone of us to come up with new plans as to how to spend our leisure hours. Therefore, the purpose of this paper is to construct an intelligent tour system on the Internet, through which users can gather all kinds of information about traveling and plan a suitable schedule to meet their requirements. The system uses web mining and hyperlink techniques to gather traveling-related and applies the minimum distance algorithm to arrange the route of a trip.

Citation:

Tung-Shou Chen, June-Horng Shiesh, Yuan Cheng Chang, and Chi-Te Huang (2002) ,"A New Intelligent Tour System Based on Web Mining Technique", in AP - Asia Pacific Advances in Consumer Research Volume 5, eds. Ramizwick and Tu Ping, Valdosta, GA : Association for Consumer Research, Pages: 303-312.

Asia Pacific Advances in Consumer Research Volume 5, 2002      Pages 303-312

A NEW INTELLIGENT TOUR SYSTEM BASED ON WEB MINING TECHNIQUE

Tung-Shou Chen, National Taichung Institute of Technology, Taiwan

June-Horng Shiesh, National Taichung Institute of Technology, Taiwan

Yuan Cheng Chang, National Taichung Institute of Technology, Taiwan

Chi-Te Huang, Providence University, Taiwan

ABSTRACT -

Now, in Taiwan, people have two days off every weekend. Much more time can be scheduled for leisure activities. Under such circumstances, it seems necessary for everyone of us to come up with new plans as to how to spend our leisure hours. Therefore, the purpose of this paper is to construct an intelligent tour system on the Internet, through which users can gather all kinds of information about traveling and plan a suitable schedule to meet their requirements. The system uses web mining and hyperlink techniques to gather traveling-related and applies the minimum distance algorithm to arrange the route of a trip.

INTRODUCTION

Because of the fast development of the Internet, the numbers of nets and websites increase at an ever-growing speed. It becomes more and more commonplace that computer users browse on the Internet instead of looking elsewhere for varieties of information about traveling that they are in need of. However, there are currently at least hundreds of websites offering information about traveling and sightseeing, among which none seems able to perfectly satisfy the needs of every single person. Travel websites nowadays can be classified into two types. The first type includes travel websites set up by individuals. When the website master has come back from a trip, she/he world show the pictures and route directions and everything on the web pages. Although they may probably be spectacular, these web pages usually do not provide such professional travel information for users as travel agencies do. On the other hand, travel websites of the second type are run by travel agencies. Developed by businesses, these websites are capable of providing information as to domestic or foreign travels, group travels, free consultations and ticket queries etc. At the present time, lots and lots of travel agencies have their own websites, and each of these websites is usually different from the others in the content, depending on what part of the business they are trying to promote. Therefore, a net-surfer who wants to plan a good trip online has to browse through perhaps dozens of such travel-agency-built websites before a perfect match can possibly be found. Of course, another weakness of such websites is that they are so too much commercialized that the information they offer is usually more business than service.

In addition, another trouble travel website browsers have to face in that the information the travel websites can offer is usually unsystematically organized. For example, if a user wants to view a series of scenic spots, she/he may need to keep switching from website to website to gather useful information. So, what one has to do is usually look for, wait for, view, and compare among websites, which is a awful waste of time. So far, we have seen no website that can dynamically help users plan their travel routes. Therefore, in this paper, we shall propose an intelligent tour system which makes use of the web mining technique and then helps users plan their travel routes on our website by their requirements. This system will expand from the local scenic spots in the Taichung area to as far as it can reach.

The rest of this paper is organized as follows. In Section 2, we shall propose our system framework, and then, in Section 3, our experimental results will be given. Finally, in Section 4, we shall draw the conclusion.

SYSTEM FRAMEWORK

The proposed system is composed of four subsystems. Its system framework diagram is shown in Figure 1 below. The four subsystems of the website are: the website interface subsystem, the plan-of-travel subsystem, the data-in-the-gathered-web-pages subsystem, and the database subsystem. The following subsections are descriptions of the four subsystems.

Website Interface Subsystem

The function of the website interface subsystem is to make dynamic two-way communications possible between users and our system. The design principle here is to use simple operations and graph guides to help users operate this system easily.

Data-in-the-gathered-web-pages Subsystem

The data-in-the-gathered-web-pages subsystem is responsible mostly for obtaining data from other travel websites and updating the data in our website. This important subsystem is constructed in two steps as follows.

Step 1: Website filtering. There are may be as many as thousands of travel websites or more on the Internet, and each website has its own features. What we do here is choose websites that better suit our needs and analyze their formats to see if we can find the rules to follow. According to our analysis result, most source codes of web pages are in the form of table tags, and most data are embedded in the tags. In this step, we must record the addresses of the selected crebsites and save them as new text files.

Step2: Development of the gathered web pages’ data programming. First, we download the selected landscape websites’ HTML web pages. Their source codes are saved as text files, and the data locations gathered by the HTML tag syntax. Now we open the text file of the travel landscape when the gathered data programming is launched. Then the programming reads the content in the file word for word. Once it gathers<td>and</td>of<Table>, it proceeds to separate them into data and tag syntax. The same thing goes on and on until the content of the file is all gone through. Then the gathered data are saved in a new file so that users can store them in databases which convenience. For example, part of the source codes of "Sale plane tickets- the path of Northeast Asia web page" is shown in Table 1 right below.

The result of data gathering is shown in Table 2. The tag syntax has been classified during the gathering process, and therefore they do not appear in the data. Finally the gathered data are stored in the database. Thus, the tasks of searching websites, gathering data, and updating the database are all done here.

2.3 Scenic Spots Database Subsystem

The scenic spots database contains a total of four data tables. It makes it possible to plan the routes on several important bases.

2.3.1 Table of Landscapes in Taichung County and City

The table offers the data gathered from web pages everywhere of the recommended landscapes in Taichung County and City.

2.3.2 Table for Time of Stay

This table records the best length of time to stay at each spot in Taichung. For example, it is recommended that one stay for about three hours to enjoy the famous works of art in Taichung Art Museum. Here the subsystem sets the time of stay by the ample principle to settle disagreements.

FIGURE 1

SYSTEM FRAMEWORK

2.3.3 Table of Places of Arrival/Departure

This table records the five ways into/out of the Taichung area as well as the distance and the time it takes to go from each of the five places of arrival/departure to every scenic spot. The five places of arrival/departure are the Taichung Train Station, the Feng-Yuan Highway Interchange, the Dah-Ya Highway Interchange, the Chung-Kang Highway Interchange and the Wang-Tien Highway Interchange respectively. The table is shown in Figure 2.

For example, scenic spot number 4001 is only one kilometer away from the Taichung Train Station, and it takes only eight minutes to get there. However, if one wants to go from scenic spot number 4001 to the Wang-Tien Highway Interchange, there will be an eighteen-kilometer distance, which takes twenty six minutes.

2.3.4 Table of Distances and Length of Time between Scenic Spots

This table records the distance and the time it takes to go across every two scenic spots in Taichung. For example, as Figure 3 shows, the distance and time are three kilometers and five minutes between scenic spot number 4001 and scenic spot number 4006.

2.4 The plan-of-travel Subsystem

The plan-of-travel subsystem is the most important part in this system. It can assist users in planning the trip according to their demands through this system. The application process of this subsystem can be divided into four steps as follows.

Step 1: Input of variable values of travel conditions. Users can input four variable values that are the "number of days", "starting point", "starting time" and "scenic spot numbers" into our travel website. For example, suppose one wants to plan a one-day tour, starting from the Taichung Train Station at nine o’clock with the numbers of the spots to visit being 4001,4007,4014, and 4004. Then he/she has to enter the conditions as Figure 4 shows.

Step 2: Looking for the spot with the shortest distance to the starting point. In this step, the system opens the "Table of places of arrival/departure" from the database and reads the distances from each starting point to each spot. Here, we use the minimum distance algorithm to sort the distances and look for the spot with the least distance to one of the starting points. See the examples given above. The distances are respectively 1, 7, 24, 8 from the Taichung Train Station to each of spots "4001, 4007, 4014, 4004". The result is 1,7,8,24 by sorting, and therefore the sequence of the spots must be modified as "4001,4007,4004,4014". Namely, scenic spot number 4001 has the least distance to the Taichung Train Station.

Step 3: Deciding the visit order. The spots are arranged in the order 4001,4007,4004,4014 in step 2 above. Then, the system draws out the "Table for time of stay" from the database and read the distances 6,4,27 that are from 4001 to the other three spots. Then, the system uses the minimum distance algorithm to obtain the order of 4,6,27 and thus the visit order is modified as 4001,4004,4007,4014. Then the system read the distances from 4004 to the other three spots and repeats the process above. The same thing goes on until the visit order is finally determined. For this example, the final visit order is 4001,4004,4007,4014.

TABLE 1

SOURCE CODE OF A WEB PAGE

TABLE 2

THE RESULT OF DATA GATHERING

Step 4: Planning travel routes. Time is an important factor in planning travel routes. The within-the-island travel is mostly a one-day or two-day tour. The system cannot arrange complex routes within such limited time. The examples below are separately a one-day tour and a two-day tour.

Planning a one-day tour route. If the user inputs the travel schedule as a one-day tour, then the system must first figure out if the total time that all the scenic spots would more than demand one day. If it falls within the limit, then the travel route planned by our system world be directly delivered to the website interface subsystem. Otherwise, the system would send a warning message to the website interface subsystem. The flow chart is shown in Figure 5.

FIGURE 2

TABLE OF PLACES OF ARRIVAL/DEPARTURE FOR THE TAICHUNG AREA

FIGURE 3

TABLE OF DISTANCES AND LENGTHS OF TIME BETWEEN SCENIC SPOTS IN TAICHUNG

FIGURE 4

TABLE OF PLACES OF ARRIVAL/DEPARTURE

FIGURE 5

THE FLOW CHART OF PLANNING A ONE-DAY TOUR

FIGURE 6

THE FLOW CHART OF PLANNING A TWO-DAY TOUR

Notes: A-time: showing the time from the starting point to the first spot.

B-time: showing the total summation of the stay time for each spot.

C-time: showing the total time needed to go between a spot and the next.

D-time: showing the length of the period from the starting of the tour until PM 6:00 that day.

Planning a two-day tour route. It is more complex to plan a two-day tour route than to do a one-day tour. No trips planned by our system lasts for more than two days so far. Some of the spots to be visited must be scheduled on the first day, and the others the second. Suppose the first day covers three spots that are the Taichung Yat-Sen Park, the Natural Science Museum, and the Tung-Hai University, and then the second day has begin with the Tung-Hai University. The rest is similar to the planning of a one-day tour. The flow chart for this part is shown in Figure 6.

Notes: F-time: the initial value namely the total time needed for the just day tour A-time?B-time(0)?C-time(0).

B-time( ): saving the stay time at each spot.

C-time( ): saving the time intervals for the transportation from places to the next places.

Scenic( ): recording the arrangement array of the whole route.

Scenic-1( ): recording the number of places to visit on the first half of the route.

Scenic-2( ): recording the number of places to visit on the second half of the route.

If F-time is smaller than D-time, then the system will add one extra spot at a time to C-time( ) and B-time( ), making F-time stay smaller than or equal to D-time. Then, one spot is brought from Scenic( ) to Scenic-1( ). If F-time is greater than D-time, then the remaining spots in Scenic( ) will be put into Scenic-2( ). Finally, the Execution result goes back to the website interface.

EXPERIMENT RESULT AND DISCUSSION

Figure 7 is the website interface that indicates all the candidate places to visit in the whole travel area. Users can input the relevant variable values of travel through this interface. The system can provide a travel route schedule, examples of which are shown in Figure 8 and Figure 10. They are plans for a one-day tour and a two-day tour, respectively.

CONCLUSIONS

In this paper, we have proposed an intelligent on-line tour planning for the reference of tour planners. At the present time, our system can already offer practical, efficient tour route plans in the Taichung area, but that does not seem good enough. For example, in the future, we can try adding some more related information to our system such as famous local products and souvenirs, snacks, restaurants/hotels and shopping areas etc. Besides that, we can also provide on-line services such as on-line restaurant reservations, plane ticket bookings, and so on. We truthfully hope that users can obtain completed travel information before they go on a trip.

FIGURE 7

THE WEBSITE INTERFACE, WHERE A ONE-DAY TOUR IS CHOSEN

FIGURE 8

THE EXECUTION RESULT OF THE ONE-DAY TOUR

FIGURE 9

THE WEBSITE INTERFACE, WHERE A TWO-DAY TOUR IS CHOSEN

FIGURE 10

THE EXECUTION RESULT OF THE TWO-DAY TOUR

REFERENCE

http://travel.network.com.tw/main/travel/index.htm

----------------------------------------