應用APWR模型進行個人化網頁建置－以線上旅遊網站為例

當顧客需要花更多的時間從多餘資訊中尋找感興趣的項目時，將會增加顧客的時間成本，因此，顧客在網?商店購物時面?的問題，?再是可?方案?足，而是資訊超載的情況，為了解決資訊超載的問題，本研究提出了 APWR 個人化網站功能需求擷取模型，幫助顧客減少過多?必要的搜尋，進而吸引?多顧客的到訪。此模型包含了(1)模糊的初步顧客知識擷取；(2)顧客知識之彙整與辨識；(3)細部顧客知識擷取；(4)清晰的顧客知識之學習與分享四個步驟，透過顧客知識管理的概念，充分挖掘存在顧客身上的知識以了解顧客真正作購買決策的理由。在模糊的初步顧客知識擷取時，展示現有網頁功能並且發放及回收 Kano 問卷；在顧客知識之彙整與辨識時，確定分群變數並且應用自適應共振??網?將線上旅遊顧客分成「議價旅遊型」、「單獨旅遊型」、「海外旅遊型」、「新手客服旅遊型」及「同好交流旅遊型」；在細部顧客知識擷取階段時，利用 Kano 二維品質模式發掘出這五種類型顧客內心對品質真正的想法與心聲，將這些顧客所最關切的品質屬性加強，進而成功設計出符合不同群類的顧客所需要的個人化網頁功能；在清晰的顧客知識之學習與分享時，把個人化網頁回寄給顧客以確認顧客是否滿意此網頁，將五種群類顧客與不分群顧客做顧客平均滿意分數的檢定，發現分群後的顧客確實有顯著滿意此分群式個人化網頁功能的概念，最後此模型也能讓新加入的顧客能被歸入分屬群體，透過誘因使新顧客填寫 Kano 問卷以檢測其對網站功能的喜好，經最相似性的計算分析後對應至所屬群體搭配之網頁，不僅達到線上旅遊業者資訊簡化的作用並能有效針對各分群進行適當行銷策略的搭配，所以整體來說，此一新的 APWR 模型確實是可以成功被應用的。

As spending extra time searching items, which the customer is interested in, in additional information, the time cost increases. Therefore, information overloaded is the main issue while shopping in online shops. In order to solve the problem, APWR model is proposed in this study to assist to reduce unnecessary searching. This model contains four steps, which are retrieving initial information, categorizing and identifying information, collecting detailed information, and learning and sharing the information. Through the concept of Customer Knowledge Management, customers’ purchase decisions can be fully understood by investigating customers' knowledge. The function of the website was demonstrated and the questionnaires were delivered while retrieving initial information. In addition, the variables of clustering was identified and Adaptive Resonance Theory was adopted to divided customers into five groups including “Bargaining,” “Travelling Alone,” “Outbound Trip,” “Newbie needed customer service” and “Communicating with travel-lover” in categorizing and identifying initial information. As collecting details, Kano 2D Quality Model was employed to discover the real opinions of the five-group customers and the personalized websites, which fit in with customers’ needs, were designed by improving the quality and attributes about which customers care. In learning and sharing the information, the personalized websites were mailed back to customers to confirm whether they were satisfied or not. The mean of five-group customers and all customers were then tested. The results show that customers after clustering are much more satisfied with the personalized websites. Furthermore, this model can categorize the new customers into these five groups. Through giving the incentive, customers were encouraged to complete the Kano questionnaire. By the Kano questionnaires, customers’ opinions of websites can be found. After the calculation and analysis, the information can be reduced and practitioners can simulate suitable strategies for different groups of customers. Finally, the APWR Model can be successfully applied.