A Customized Visiting Route Service under RFID Environment Chieh-Yuan Tsai*, Bo-Han Lai

Department of Industrial Engineering and Management Yuan-Ze University 135 Yuan-Tung Rd.,

Chung-Li, Taoyuan 320, Taiwan *cytsai@saturn.yzu.edu.tw

Abstract—How to provide a high quality service according to consumer preference becomes a critical issue for amusement park to survive in a rapidly changing environment. To fulfill the need, this research proposes a customized visiting route service that provides tourists what facilities they should visit and in what order. In the studied environment, all regions are covered by Radio-Frequency Identification (RFID) readers so that the visiting behavior of a tourist (i.e. visiting location, sequences, and corresponding timestamps) can be collected and stored in a route database. The proposed route recommendation service consists of two major modules. The first module is to discover the frequent Location-Item-Time (LIT) sequential patterns using the proposed sequential pattern mining procedure. In the second module, the route suggestion procedure will filter the LIT sequential patterns under the constraints of intended-visiting time, favorite regions with its related visiting time, and favorite recreation facilities, then select the top-k suggested routes to guide the visitors. To show the feasibility of the proposed route recommendation system, the Tokyo DisneySea in Japan is used as an example. Based on the experimental results, it is clear that the recommended route can not only follow previous tourists’ visiting experiences but also satisfy the visitor’s customized requirement.

Keywords—recommendation service; visiting route suggestion; RFID; sequential pattern mining; amusement park

I. INTRODUCTION Customized tourism services aim at helping users to

find what they are looking for by comparing the user profile to some reference characteristics without spending much time and effort [1]. Therefore, a variety of approaches have been used to perform recommendations in these domains, including content-based, collaborative filtering and hybrid approaches [2-6]. Most of the mentioned studies focused on recommending products and services rather than how to offer tourists a customized visiting itinerary that guides them completely through their trip. To fill this gap, [7] took previous popular visiting behaviors as the foundation and developed a sequential pattern based route suggestion system to generate personalized tours. [8] developed a route recommendation system that provides personalized visiting routes for tourist in theme parks that consider a set of visiting sequences. Obviously, those researches provide the concept of providing valuable visiting sequence to help tourists complete their trips on time. Without a personalized route suggestion, tourists tend to make an inefficient trip or even get lost in the complex them park environment. However, these recommendation systems did not take the geographic constraints of environment into consideration. That is,

these recommendable facilities might not be neighboring. On the other hand, the visitor can’t understand which region those facilities belongs to.

In recent years, there has been a dramatic increase in wireless communication and location-aware technologies. These wireless communication technologies such as GPS (Global Positioning System), RFID (Radio Frequent Identification) and WLAN (Wireless Local Area Network) can be used to collect location data of moving objects. The movements of an object are simply described as a sequence of spatial locations and temporal properties [9-12]. Actually, some researchers have developed sequential pattern mining algorithm related to location and items. [13] proposed a Conditional Sequential Base mining (CSB-mine) algorithm. The CSB-mine algorithm, which is based directly on conditional sequence bases of each frequent event, is used to improve the design of web sites by analyzing user behaviors for personalized services. [14] developed a mining mobile sequential patterns algorithm to better reflect the customer usage patterns in the mobile commerce environment, which takes both the moving patterns (Location) and purchase patterns (Items) of customers into consideration. [11] proposed a Sequential mining of Moving Patterns based on spatial constraints in Mobile environment (SMPM) algorithm. The SMPM algorithm has a property of location constraints which can avoid mining the same projected database. However, these algorithms did not take the time between items (Time) into consideration and are hard to be used in a personalized route recommendation service.

To solve the above difficulties, this research develops a Location-Item-Time sequential mining procedure that provides customized visiting routes for tourists in the amusement park when considering a set of visiting constraints. This paper is organized as follows. Section 2 formally defines the research problem and proposes the architecture of route recommendation system for amusement park. Section 3 provides an implementation case to demonstrate the feasibility of the proposed method. Section 4 summarizes the conclusions and points out possible future works.

II. RESEARCH METHOD

A. Environment Assumption and System Overview Typically, an amusement park is divided into several

regions and each region might contain a set of recreation facilities. In this study, it is assumed that each region is covered by RFID readers. In addition, RFID readers are

2013 IEEE 37th Annual Computer Software and Applications Conference Workshops

978-0-7695-4987-3/13 $26.00 © 2013 IEEE

DOI 10.1109/COMPSACW.2013.67

397

assumed to be installed in the entrance of each recreation facility, and entrance and exit of the park. When a visitor with a RFID tagged wristband enters a region or facility, RFID readers will record the RFID tag code, region id, facility id, and the time to of entering the region to the route database. This recording process continues until the visitor leaves the amusement park. Let’s take the layout in Figure 1 to explain the recording process. At timestamp t1, a visitor passes the entrance k11 of the park in region B. Then, she moves to region A at timestamp t2, region F at timestamp t3, and region G at timestamp t4. In region G, she takes facility k1. After that, she moves to region K at timestamp t5, region O at timestamp t6. In region O, she takes facilities k2 and k3. The recording process continues until the visitor leaves the park from the exit k12 of the park in region B. Finally, the route sequence <(B, t1, {k11}), (A, t2, φ ), (F, t3, φ ), (G, t4, {k1}), (K, t5, φ ), (O, t6, {k2, k3}), (K, t7, φ ), (G, t8, φ ), (B, t9, {k12})> will be collected and stored in the route database.

Fig. 1. An illustrative example for route sequence generation.

Whenever a visitor wants to request a route recommendation, he/she can reach the kiosk machine in food courts, shops and information centers and input his/her preference to the route recommendation system in the machine. The preference might include intended total visiting time, favorite regions, intended visiting time in the favorite regions, and favorite recreation facilities. The proposed route recommendation system consists of two major modules. The first module is to generate a set of frequent Location-Item-Time (LIT) sequential patterns from the route database, and the second module is to retrieve suitable LIT sequential patterns based on a visitor’s preferences. The first module can be further divided into two procedures. The first procedure is to preprocess the route sequences in the route database so that unsuitable routes are modified and deleted. The second procedure is to discover the LIT sequential patterns using the proposed Location-Item-Time sequential pattern mining procedure. In the second module, the route recommendation procedure will evaluate the similarity between the visitor’s preference and candidate LIT routes, then retrieve suitable routes for the visitor. The more detail a visitor provides his/her preference, the more satisfied suggestion he/she receives.

B. Location-Item-Time (LIT) Sequential Patterns Let N = {n1, n2, …, ng} be the set of cells (regions) in

the amusement park and K= {k1, k2, …, kh} be the set of

items (facilities, entrance, or exit). In the route database RD, a record is represented by <sid, rs> where sid is the identifier of the record and rs is a route sequence. Formally, rs is represented as <(B1, t1, itemset1), (B2, t2, itemset2), …, (Bn, tn, itemsetn)> where (Bi, ti, itemseti) is an event; Bi is the visited region and Bi � N; ti stands for the timestamp that region Bi is first entered and ti-1�ti for

ni ≤≤2 ; itemseti is the set of items visited in region Bi and itemseti ⊆ K. Without timestamp information, <Bi, itemseti> is called a transaction if itemseti is a non-empty set. A transaction pattern is defined as <Bi; z> where z is the non-empty subset of itemseti. If the length of z is k, <Bi; z> is called a k-transaction pattern.

Let ii ttt −=Δ +1 be the time interval between two successive events where ni ≤≤1 and Tc be a set of given constants for rc ≤≤1 . Then, the time interval tΔ can be transferred as one of elements in the set of discrete time intervals TI = {I1, I2, …, Ir} by

���

≤<≤Δ<≤Δ<

=Δ− rjTtTI

TtItDiscTI

jj 1forif0if

)(1

11 (1)

For example, assume T1 = 10, T2 = 20, T3 = 30, T4 = 40, T5 = 50, and T6 = 60. Therefore, the set of discrete time interval be TI = {I1, I2, I3, I4, I5, I6}, where I1:

0< tΔ ≤ 10, I2:

10< tΔ ≤ 20, I3: 20< tΔ ≤ 30, I4:

30< tΔ ≤ 40, I5: 40< tΔ ≤ 50, I6: 50< tΔ ≤ 60.

Definition 1. Let �={�1, �2, …, �n} be the set of transaction patterns and TI = {I1, I2, …, Ir} be the set of discrete time intervals. A sequence � = (D1, �1, D2, �2,…, Dq-1, �q-1, Dq) is a Location-Item-Time (LIT) sequence if �Ds ∈ for

qs ≤≤1 and TIs ∈ε for 11 −≤≤ qs .

C. Location-Item-Time Mining Procedure Similar to the work of [14], the proposed method for

mining Location-Item-Time (LIT) sequential patterns consists of three phases which are the large-transaction generation phase, large-transaction transformation phase, and location-Item-Time sequential pattern generation phase.

1) Large-transaction generation phase The large-transaction generation phase is to determine

the large transactions from the route database RD. This phase consists of two steps. The first step applies Apriori algorithm for RD to find the set of all large transaction patterns. If the support count of a k-transaction pattern is greater than or equal to the user-specified minimum support count (called min_sup_count), the k-transaction pattern will be called a large k-transaction pattern. In the second step, the set of large itemsets is mapped to a set of unique symbol for reducing the later computation time.

2) Large-transaction transformation phase The large-transaction transformation phase transforms

route sequences into the maximal large-transaction sequences. This phase repeatedly determine which parts of a given set of large 1-sequential patterns appear in route sequences. Through the phase, the record with the form of

398

<sid, rs> in route database RD will be transferred to the form of <sid, maximal large-transaction sequence, path> in transformed route database TRD.

3) Location-Item-Time sequential pattern generation phase

Next, a Location-Item-Time (LIT) sequential pattern algorithm is developed to generate all large LIT sequential patterns from TRD. Similar to [15], the proposed LIT sequential pattern algorithm is based on PrefixSpan mining concept.

Definition 2. For a maximal large-transaction sequence � = (<(B1; z1), t1>,<(B2; z2), t2>,…, <(Bn; zn), tn>) and a Location-Item-Time (LIT) sequence � = (D1, �1, D2, �2,…, Dq-1, �q-1, Dq), � is said to be contained in � or � is a LIT subsequence of � if the integers njjj q ≤<<<≤ ...1 21 exist such that,

1. D1= );(11 jj zB , D2= );(

22 jj zB , …, Dq= );(qq jj zB .

2. 1−

−ii jj tt satisfies the condition of �i-1 for qi ≤≤2 .

Definition 3. support_countTRD(�) = |{(sid, maximal large-transaction sequence, path) | (sid, maximal large-transaction sequence, path) ∈ TRD ∧ � is contained in TRD}|. A LIT sequence � is called a LIT sequential pattern if the percentage of records in TRD consisting of � is greater than or equal to the pre-defined minimum support, called min_sup. Simply stated, � is named a LIT sequential pattern in TRD if support_countTRD(�) ≥ |TRD|× min_sup. A LIT sequence whose length is l is denoted as a l-LIT sequence.

In order to simplify the computation, a symbol is used to arrange these postfixes. The �-projected database defined by the collection of postfixes of maximal large-transaction sequences in TRD with respect to � is denoted as TRD|�. A table LIT_Table is used to store this type of relation, where a column corresponds to a large-transaction pattern and a row corresponds to a time-interval in TI = {I1, I2, …, Ir}. Each cell LIT_Table(Ii,

'iγ ) in the table

records the number of transactions in TRD|� which contains transaction pattern and the time difference between this transaction pattern and the last transaction pattern of � lies within Ii. Processing every transaction in TRD|� sequentially enables LIT_Table to be formed and the frequent cells to be identified. If the cell LIT_Table(Ii,

'iγ ) is a frequent cell, (Ii, '

iγ ) can be appended to � to yield a LIT sequential pattern 'α , and to construct the 'α -projected database TRD 'α . Recursively discovering the LIT sequential patterns in TRD 'α finally yields all LIT sequential patterns in TRD.

D. Route Recommendation Procedure When a visitor requires a route suggestion, he/she

simply enters preference to the route recommendation system in the kiosk. The visitor’s preference can be represented as a VP vector:

VP=<ITVT, <FR1,FItems1,IRVT1>,

<FR2,FItems2,RVT2>, …> (2)

where ITVT is the intended total visiting time. FRi is the favorite region i , FItemsi is the set of favorite facilities in FRi, and IRVTi is the intended visiting time in FRi. Note that the more information a visitor enter, the more satisfied suggestion the visitor can obtain. For example, VP = <420, <G, {k1}, 90>, <O, {k2, k3}, 120>> indicates that the visitor intends to spend 420 minutes in the amusement park. In addition, he/she would like to spend 90 minutes in region G and take recreation facility k1 in region G, and 120 minutes in region O and take recreation facility k1 and k3 in region O.

The number of LIT sequential patterns generated from the proposed algorithm might large. However, not all patterns are considered as candidate LIT routes and will be further examined except a pattern satisfies the following two rules. First, if a LIT sequential pattern does not contain entrance and exit, the pattern will be removed. Second, if a LIT sequential pattern does not satisfy the time constraint provided by the visitor, the pattern will be removed. The time constraint is defined as follows.

As mentioned before, the time interval tΔ can be transferred as one of elements in the set of discrete time intervals TI = {I1, I2, …, Ir} according to Equation (1). Let a LIT sequential pattern � be represented as (D1, �1, D2, �2,…, Dq-1, �q-1, Dq). The total visiting time of � can be represented as VT�

= ( βLBVT , β

UBVT ] where the lower bound

of VT� is defined as � −

== 1

1)(q

s sLBLB fVT εβ and the upper

bound of VT� is defined as � −

== 1

1)(q

s sUBUB fVT εβ . If

ITVTVTLB ≤β and ITVTVTUB ≥β , we say that LIT sequential pattern � satisfies the a visitor’s time constraint where ITVT is the visitor’s the intended total visiting time in Equation (2).

The similarity measure between VP=<ITVT, <FR1, FItems1, IRVT1>, <FR2, FItems2, IRVT2>,…> and candidate LIT route � = ((B1; z1), �1, (B2; z2), �2,…, (Bq-1; zq-1), �q-1, (Bq; zq)) is designed based on the following concepts. First, the intended visiting time for region i, IRVTi, in VP will be mapped as one of the elements in TI = {I1, I2, …, Ir} according to Equation (1) for all i. Second, when conducting the similarity evaluation, <FRi, FItemsi, IRVTi> in VP and <(Bj; zj), �j> in � are considered as comparison units. Third, if FRi and Bj are the same region, similarity evaluation between <FItemsi, IRVTi> and <zj, �j> will be initialized. Base on above concepts, the similarity between ith unit in VP and the jth unit in � is defined as:

ji

jiji BFR

BFRjiTSimwjiISimwwSim

≠=

��� ×+×+×

=ifif

0),(),(1 321

,

(3)

where w1, w2, and w3 are the important degrees for region, facility, time-interval considerations respectively, and w1 + w2 + w3 = 1. ISim(i, j) is the itemset similarity between FItemsi and zj which is defined as:

399

||/||),( iji FItemszFItemsjiISim ∩= (4)

where ∩ is the set union operator and | | is the cardinality of the set. In addition, TimeIntervalSim(i, j) is the time interval similarity between IRVTi and �j which is defined as:

)(/|)()(|1),( rji IffIRVTfjiTSim ε−−= (5)

where || ⋅ is the absolute value operator and f (Ib) is the rank of the time-interval Ib in TI and is defined as f (Ib) = b where b = 1, …,r. The similarity between VP and � is defined as:

||),(),(||

1

||

1VPjisimVPSim

VP

i j��

= ==

β

β (6)

where || ⋅ is the length of the sequence. After the similarities between VP and all candidate routes are derived, they are sorted in decreasing order. The candidate routes with top three small values are returned back to the kiosk machine as suggested routes. It is possible that more than one candidate routes have the same similarity. In this case, the route having larger number of total facilities will have higher rank.

III. IMPLEMENTATION AND EXPERIMENT RESULTS

A. Case Description and Route Generator In this study, a simplified layout of Tokyo DisneySea

in Japan is used as an example to illustrate the feasibility of the proposed route recommendation system. As shown in the figure, there are seven thematic regions and thirty-four recreation facilities (k1 to k34). The entrance (k35) and exit (k36) of the amusement park are located at region A. The operation time of the amusement park is from 9:00 a.m. to 10:00 p.m.. According to the tourism reports, five must-visited recreation facilities and seven popular facilities are summarized in Table 8.

Fig. 2. The simplified layout of the implementation example

To simulate visiting behaviors, a route generator is developed. In the generator, visitors start their routes from the entrance and finish at the exit. The regions that visitors

pass through must be adjacent. Table 9 shows parts of the route sequences generated by the route generator.

TABLE I. A SIMULATED ROUTE DATABASE

Sid Route sequence (min.)

1 <(A,7,{k35}),(B,17,{k1}),(C,126,{k4,k7}),(F,231,{k22}),

(C,327,�),(B,343, �),(A,363,{k36})>

2 <(A,9,{k35}),(B,22, �),(C,45,�),(F,75,{k23,k24}),

(G,222,{k25,k26}),(D,363, �),(B,378, �),(A,405,{k36})>

3 <(A,8,{k35}),(B,16,{k1}),(C,98,{k4,k9,k10}),(B,328, �),(A,351,{k36})>

4 <(A,8,{k35}),(B,17, �),(E,43, �),(H,73,{k33,k34}),(E,172, �),(D,201,{k12,

k13}),(F,365,{k22,k24}),(C,524, �),(B,554,{k3}),(D,620, �),(G,636,

{k25}),(D,690, �),(B,719, �),(A,737,{k36})>

5 <(A,2,{k35}),(B,9,{k1,k2,k3}),(D,221,{k12,k13}),(C,371,{k5,k6,k8,k9}),

(B,666, �),(A,688,{k36})>

� �

B. Route Recommendation Generation 1) Location-Item-Time mining procedure module

Next, the proposed Location-Item-Time sequential pattern mining procedure is applied to discover the frequent Location-Item-Time (LIT) sequential patterns. Before executing the algorithm, the first task is to determine the minimum support and the set of discrete time intervals. For simplicity, the time intervals in this study are set as equal length of 30 minutes and the minimum support is set as 0.02%. Therefore, the set of discrete time intervals are TI = {I1, I2, I3, …, I20}, where I1: 0 30t< ≤ , I2: 30 60t< ≤ , I3: 60 90t< ≤ , …, I20: 800760 ≤< t . Based on the settings, The mining procedure discovered 380,735 LIT sequential patterns.

2) Route recommendation generation module Assume a new user intends to spend 420 minutes in the

amusement park and wishes to play recreation facility {k12} of region D, and recreation facility {k22} of region F. In addition, he/she wishes to spend 150 minutes in region D and 120 minutes in region F respectively. Thus, the user preference UP vector can be formed as <420, <D, {k12}, 150>, <F, {k22}, 120>>. In addition, the important degrees for region w1, facility w2, time-interval w3 in Equation (3) are set equally as 1/3. Based on the set of discrete time-intervals I, the total visiting time (VTu) of each LIT sequential pattern can be calculated. After deleting the sequential patterns that do not contains entrance and exit as well as the patterns that do not satisfy the time constraint, 5,471 candidate LIT routes can be found. Table 11 shows the ranking of candidate LIT routes, and Table 12 shows the path information of candidate LIT routes. Finally, the route recommendation generation module will suggest the first three ranked candidate LIT routes to the new user.

TABLE II. THE PATH INFORMATION OF EACH CANDIDATE LIT ROUTES

Ranking Candidate LIT route Sup.

1 <A;k35>,I1,<B;k2>,I4,<D;k12,k13>,I4,<F;k22>,I4,<A;k36> 5

2 <A;k35>,I1,<B;k2>,I4,<D;k12>,I4,<F;k22>,I4,<A;k36> 5

3 <A;k35>,I1,<B;k2>,I3,<D;k12,k13>,I4,<F;k22>,I5,<A;k36> 5

400

Figure 10 shows visiting path of the candidate LIT route with number one ranking. The recommendation system suggests the visitor starts the trip from the entrance in region A. Within 30 minutes (time-interval I1), the visitor is suggested to take Venetian Gondolas (k2) recreation facility in region B. After 120 minutes to 160 minutes (time-interval I4), the system suggests the visitor takes Journey to the Center of Earth (k12) and 20,000 Leagues Under the Sea (k13) in region D. Again, after 120 minutes to 160 minutes (time-interval I4), the visitor is suggested to take StormRider (k22) in region F. Finally, after 120 minutes to 160 minutes (time-interval I4), the visitor is suggested to leave the Tokyo DisneySea amusement park from the exit in region A by passing through regions C, B, and A sequentially.

Fig. 3. The visiting sequence recommendation based on the user’s preference

Table 18 summarizes the execution time of each phase of mining procedure module and route recommendation generation module respectively. In the real situation, the first module is performed whenever. On the other hand, the second module is executed whenever there is a request from a visitor. It is clear that, when the number of route sequences increases, the execution time for the mining procedure module increases linearly. The execution time for the route recommendation generation module increases slowly than the mining procedure module. Although the mining procedure module takes much time to execute, the module is conducted off-line. That is, this module will be conducted daily or weekly. On the other hand, the average execution time for the route recommendation generation module is within 778.27 second.

TABLE III. THE EXECUTION TIME (IN SECOND) OF EACH PHASE OF MINING PROCEDURE MODULE AND ROUTE RECOMMENDATION

GENERATION MODULE

The number of route sequences

10,000 13,000 16,000 19,000 22,000 26,000

Sequential

pattern mining

procedure

module

8241.6 7243.0 10565.8 11090.9 15223.8 17288.83

Route

recommendation

generation

1215.6 488.4 809.3 580.0 824.8 751.46

module

IV. CONCLUSIONS In order to provide a customized route suggestion,

many recommendation systems have been developed in the past decade. Most of the mentioned studies focused on recommending products and services rather than how to offer tourists a customized visiting itinerary that guides them complete their trip. Without a customized route suggestion, tourists tend to make an inefficient trip or even get lost in the complex them park environment. To bridge the gap, this research takes location, items, and time factors into consideration at the same time and develops a new sequential pattern mining procedure that provides customized visiting routes for tourists in the amusement park when considering a set of visiting constraints. The experimental results show that the suggested routes not only can satisfy the visitor’s constraints, but also can suggest more recreation facilities within the visitor’s intended-time.

There are some directions to improve the proposed route recommendation system in the future. First, the minimum support, the length of time interval and the important degree are decided by managers. It worthwhile to automate this decision procedure by adopting optimization techniques when achieving a practical objective function of route recommendation system. Second, the proposed route recommendation generation does not record the current location that the visitor makes a suggestion request, so that the suggested route might not begin at the current location when visitor makes request. To overcome this drawback, the future study should consider the record of visitor’s current location in route recommendation generation. Last, in real visiting situation, the visitor may play different recreation facilities on his/her return trip. However, the proposed similarity measurement of route recommendation generation cannot handle this situation. To overcome this drawback, the future study should involve this situation.

REFERENCES [1] Kabassi, K., “Personalizing recommendation for tourists,”

Telemetric and Informatics, 27(1), pp. 51-66, 2010. [2] Albadvi, A., and Shabbazi M., “A hybrid recommendation

technique based on product category attributes,” Expert Systems with Application, 36(9), pp. 11480-11488, 2009.

[3] Cho, Y.-B., Cho, Y.-H., and kim, S.-H., “Mining changes in customer buying behavior for collaborative recommendations,” Expert Systems with Applications, 28(2), pp. 359-369, 2005.

[4] Liu, D., and Chang, M., “Recommend touring routes to travelers according to their sequential wandering behaviors,” Proceeding of the 10th International Symposium on Pervasive Systems, Algorithms, and Networks, pp. 350-355, 2009.

[5] Tan, X., Yao, M., and Xu, M., “An effective technique for personalization recommendation based on access sequential patterns,” Proceeding of 2006 IEEE Asia-Pacific Conference on Services Computing, pp. 42-46, 2006.

[6] Oku, k., Nakajima, S., Miyazaki, J., Uemura, S., and kato, H., “A recommendation method considering users’ time series contexts,” Proceedings of the 3rd International Conference on Ubiquitous Information Management and Communication, pp. 465-470, 2009.

[7] Tsai, C.-Y., and Lo, P.-H, “A sequential pattern based route suggestion system,” International Journal of Innovative Computing, Information and Control, 6(10), pp. 4389-4408, 2010.

401

[8] Tsai, C.-Y., and Chung, S.-H., “A personalized route recommendation service for theme parks using RFID information and tourist behavior,” Decision Support Systems, 52(2), pp. 514-527, 2012.

[9] Lee, Y.-S., and Yen S.-J., “Incremental and interactive mining of web traversal patterns,” Information Sciences, 178(2), pp. 287-306, 2008.

[10] Yavas, G., katsaros, D., Ulusoy, O., and Manolopoulos, Y., “A data mining approach for location prediction in mobile environments,” Data and knowledge Engineering, 54(2), pp. 121-146, 2005.

[11] Zou, Y.-G., and Yu, H., “Moving sequential pattern mining based on spatial constraints in mobile environment,” Proceeding of the 2010 IEEE International Conference on Intelligent Computing and Intelligent System, 2, pp. 103-107, 2010.

[12] Zhou, B., Hui, S.C., and Chang, k., “An intelligent recommender system using sequential web access patterns,” Proceeding of the 2004 IEEE Conference on Cybernetics and Intelligent Systems, pp. 393-398, 2004.

[13] Zhou, B., Hui, S.C., and Fong, A.C.M., “Efficient sequential access pattern mining for web recommendations,” International Journal of knowledge-based and Intelligent Engineering Systems, 10(2), pp. 155-168, 2006.

[14] Yun, C.-H., and Chen, M.-S., “Mining mobile sequential patterns in a mobile commerce environment,” IEEE Transactions On Systems, Man and Cybernetics Part C: Application and Reviews, 37(2), pp. 278-295, 2007.

[15] Chen, Y.-L, and Chiang, M.-C., and ko, M.-T., “Discovering time-interval sequential patterns in sequence database,” Expert Systems with Applications, 25(3), pp. 343-354, 2003.

402