2017 Asia-Pacific Egieerig ad Techology Coferece (APETC 2017) ISBN: 978-1-60595-443-1 Otology-based Decisio Support System with Aalytic Hierarchy Process for Tour Pacage Selectio Tie-We Sug, Chia-Jug Lee, Pei-Wei Tsai, Sheg-Hui Meg ad Fu-Tia Li ABSTRACT The popularity of Iteret ad growig B2C electroic commerce owadays mae product or service iformatio easy to be acquired. However, maig a optimal choice from the various alterative products becomes a laborious process. I this paper, a otology-based Decisio Support System (DSS) with Aalytic Hierarchy Process (AHP) was proposed for the specific applicatio of tour pacage selectio. The system is composed of two subsystems, the product gatherer ad the decisio maer, which are used to fid out right products ad mae a expected choice respectively. I the product gatherer subsystem, a otology-based web service architecture with Web Otology Laguage (OWL) was established for the sematic cotet processig of product iformatio. The Simple Object Access Protocol (SOAP) is utilized to establish the commuicatio iterface ad gather XML-based cotets through Remote Procedure Calls (RPC) betwee the system ad the database servers of travel agecies. I the decisio maer subsystem, the Aalytic Hierarchy Process is utilized to mae a optimal decisio for satisfyig the requiremet give by the cosumer. The system aims to help cosumers to avoid fallig ito decisio-maig hesitatio ad get a expected choice from various ad similar products. INTRODUCTION The advace of Iteret ad WWW techologies has made a great impact o the huma life ad busiess. It has caused the electroic commerce applicatios to grow extraordiarily, especially for the busiess-to-cosumer applicatios. However, the coveiece also mae various ad similar products difficult to be chose by cosumers. Maig decisio with may tradeoff cosideratios amog products is the major cause of such a hesitatio. To solve the problem, it requires a itelliget decisio-maig process. This paper proposes a architecture ad a itelliget Decisio Support System (DSS)[1] to help cosumers mae the purchase Tie-We Sug 1,*, Chia-Jug Lee 1, Pei-Wei Tsai 1, Sheg-Hui Meg 1, ad Fu-Tia Li 2 1 Fujia Provicial Key Laboratory of Big Data Miig ad Applicatios, College of Iformatio Sciece ad Egieerig, Fujia Uiversity of Techology, Chia 2 Departmet of Electrical Egieerig, Tug Fag Desig Uiversity, Taiwa 309
decisio by coductig the Web Service ad Aalytic Hierarchy Process (AHP)[2]. There are two fudametal compoets, the product gatherer ad the decisio maer, i the system. The first oe is majorly resposible to complete tourist iformatio aggregatio from distributed database servers offered by the travel agecies. The other oe acts as a decisio-maig assistat. SYSTEM ARCHITECTURE The proposed system is a 3-tier architecture, as show i Fig. 1, which cotais the frot-ed Cliet Tier (Web Browser), middle Applicatio Tier (Web Applicatio) ad bac-ed Data Tier (Web Service). Cosumers who wat to tae a tour could evaluate some tour pacages before maig the purchases. The system ca help users to collect tour pacages iformatio from distributed database servers provided by differet travel agecies ad recommed a optimal oe accordig to user prefereces. For the purpose of deliverig data via a stadard commuicatio iterface, the system utilizes Simple Object Access Protocol (SOAP)[3] for the data commuicatio betwee middle tier ad bac-ed tier. This will require the employ of database system which supports Web Service fuctioality. Microsoft SQL Server has built-i ative XML Web Service features with SOAP i its egie ad is utilized i the proposed system. O the other had, data must be expressed i a structured ad stadard maer for iteroperability; therefore the system uses Web Otology Laguage (OWL)[4] to defie the XML-based otology about tour iformatio. Tour pacage data will be stored ad delivered i a XML formatio. Figure 1. System Architecture. PRODUCT GATHERER SUBSYSTEM The product gatherer subsystem was divided ito two parts: data requester ad data provider. The data requester is resposible to request the data provider to deliver tour pacage data which satisfies the cosumer s prefereces. The preferece settig will be fiished i the user iterface. After receivig the request, data provider will query its database ad retur the results. Data requests ad 310
resposes operate uder the Web Service-based eviromet; thus, data requester acquires required data via a Remote Procedure Call (RPC) to the data provider. To do this, the XML ad HTTP-based SOAP stadard protocol is used. User s preferece settig ad retured tour pacage data should be ecapsulated respectively i the SOAP etwor pacets of request ad respose. The Otology sematic techique[5] is also used i the ecapsulatio; thus, Otology-based XML schema is used i the expressio of preferece settig ad retured data. The OWL laguage was utilized to pre-defie the travel otology i the system implemetatio. The used RPC performs the detail data query process. It is implemeted by the database Stored Procedures which were created by the CREATE PROCEDURE statemet ad saved as a collectio of Trasact-SQL statemets i the database system. Moreover, to set up the database egie as a Web Service provider that ca liste for SOAP requests, it requires the creatio of a HTTP Edpoit beforehad. A HTTP Edpoit is created for use with SQL Server to liste ad receive requests o a TCP port (Ex: port 80) ad to start up the executio of idicated stored procedures. The edpoit was created by the CREATE ENDPOINT statemet i the database system. A additioal remar is that oly oe data requester compoet was built i the middle-tier AHP-DSS system, but a data provider compoet should exist i each system owed by every joied travel agecy. I other words, AHP-DSS system ca sed SOAP requests to may HTTP Edpoits distributed i respective database servers that provided by the travel agecies ad ca get various tour pacages data from these agecies for further tour pacage selectio decisio. I curret stage of the proposed system, the list of joied agecies was recorded withi a table i the AHP-DSS system. I the future wor stage, the Web Service registratio mechaism UDDI (Uiversal Descriptio, Discovery ad Itegratio)[6] will be coducted ito the system. DECISION MAKER SUBSYSTEM Oce the data of various tour pacages has bee acquired, the ext step should help the user to mae a preferred decisio. Maig decisio by users with may tradeoff cosideratios is a hesitatig process. Aalytic Hierarchy Process is used i the system to solve such a problem. It is a structured techique for assistig people to mae complex decisios ad ca be divided ito five steps: (1) model the problem as a decisio hierarchy; (2) calculate local weight vectors; (3) chec cosistecy; (4) calculate global weight vector; ad (5) mae the decisio. The first step of AHP is to costruct a decisio hierarchy which the top level is the decisio goal; the secod level is the collectio of cosideratio factors, ad the third level is composed of differet tour pacages. Fig. 2 is a illustratio of a costructed hierarchy. 311
Figure 2. Decisio Hierarchy. The secod step is to calculate local weight vectors, which are the weight vector of decisio factors to the goal ad the weight vector of products to each factor. Both of the weight vectors ca be calculated with the same procedure. I this step, firstly it requires to mae a pairwise compariso matrix through pairwisely comparig each decisio factor by usig judgmets to determie the relative importace of oe factor over aother. The pairwise comparisos should be judged by the user with scales 1 to 9. The matrix is show i Equatio (1), where a deotes the weight ratio of the i th factor to the j th factor, ad is the umber of factors. Equatio (2) shows the local weight vector of decisio factors to the goal, ad a eigevector W ca be obtaied from the eigevalue equatio show i Equatio (3). The Average of Normalized Colums (ANC) method show i Equatio (4) also ca be used to obtai a approximate value of the eigevector W. This vector idicates the weights of each decisio factor to the goal. A a W a11 a12 a13... a1 a21 a22 a23... a2 a a a 31 32...... 3............... a1 a2 a3... a w1 w2 w 3, 1, 2, 3,..., 1... w w (2) j1 a w w, i 1, 2, 3,..., (3) max 1 wi a aj, i 1, 2, 3,..., (4) j1 1 Sice user s judgmets may cause to a icosistet pairwise compariso matrix ad a sigificat icosistecy will lead to a uacceptable decisio result, the ext step is to chec the cosistecy of the matrix. If the matrix satisfies the followig Equatios (5) ad (6), it is a cosistecy matrix. a ji (1) a 1/ (5) a i a a (6) j 312
After the calculatio of level-2 weight vector, the ext step is to determie the level-3 weight vector, that is, the weight vector of products to each factor. The procedure is the same as described above. There will be seve pairwise compariso matrices i the case sice there are seve factors cosidered. Fially, the global weight vector ca be obtaied by the Equatio (7), where t i is the global weight of i th product, b is the local weight of i th product to j th factor, ad w j is the local weight of j th factor. The product with largest value is the preferred ad optimal choice for the decisio. Fig. 3 shows the example that tour pacage E is the optimal choice. t i b w j j1 (7) CONCLUSION Figure 3. Decisio result (a example). This paper proposes a architecture that itegrates otology-based sematic web service with Aalytic Hierarchy Process to provide cosumers the decisio support assistace ad recommedatio for tour pacage selectio. Sice maig decisio with may tradeoff cosideratios usually causes cosumers fallig ito a hesitatio, the proposed itelliget system that composed of subsystems of product gatherer ad decisio maer ca help to solve this situatio ad offer a preferred ad optimal choice to the cosumer for further purchase. The system ca be improved i the future by itroducig UDDI service registratio mechaism ito data gatherer subsystem. This ca improve the process flexibility for tourist data aggregatio from travel agecies. This wor is partially supported by Fujia Provicial Departmet of Educatio, Chia, uder Grat No. FJJKHX16-031. REFERENCES [1] D.J. Power, Decisio Support, Aalytics, ad Busiess Itelligece (Busiess Expert Press, New Yor, 2013). [2] M. Bruelli, Itroductio to the Aalytic Hierarchy Process (Spriger, New Yor, 2015). [3] K. Kadouh, K. Albashiri, Improvemet of Data Trasfer over Simple Object Access Protocol (SOAP), It. J. Comput. Electr. Auto. Cotrol. If. Eg., 8(2), 286-289 (2014). [4] S.T. Cao, L.A. Nguye, A. Szalas, The Web Otology Rule Laguage OWL 2RL+ ad Its Extesios, Lect. Notes Comput. Sci., 8342, 152-175 (2014). [5] V. Jai, M. Sigh, Otology Based Iformatio Retrieval i Sematic Web: A Survey, It. J. If. Techol. Comput. Sci., 5(10), 62-69 (2013). [6] Y. Dai, Y. Feg, Y. Zhao, Y. Huag, A method of UDDI service subscriptio implemetatio, Proceedigs of 5th IEEE Iteratioal Coferece o Software Egieerig ad Service Sciece (IEEE, Beig, Chia, 2014). 313