透過您的圖書館登入
IP:3.16.147.124
  • 學位論文

以價值共創觀點探索LDA主題模型在線上評論的應用–餐飲業為例

Value co-creation perspective on exploring the application of online consumer reviews based on LDA : A case study of Catering Industry

指導教授 : 郭瑞祥
若您是本文的作者,可授權文章由華藝線上圖書館中協助推廣。

摘要


過去餐飲業評估顧客服務體驗與評價時,最常被使用的方法即為設計內部「用餐滿意度問卷」,以顧客被動填寫滿意度方式進行服務品質的衡量。然而,此種方式僅能提供靜態的資訊蒐集,難以瞭解顧客在服務過程中的動態體驗與回饋。近年顧客體驗當道,企業經營從產品主導的邏輯,轉向服務和顧客主導的邏輯(Service-Dominant Logic, S-D Logic)演進的現象,其主張以顧客為中心,強調企業可以與顧客價值共創(Lusch Vargo, 2006; Lusch, Vargo, Tanniru, 2010; Vargo Lusch, 2016; Vargo, Maglio, Akaka, 2008),在顧客服務網路中,藉由彼此間的互動及資源整合的活動中所實現。 本研究透過此觀點,我們以數據挖掘(Data mining)方法,了解線上評論消費者真實的聲音 (Voice of the Customer, VOC),並使用隱含狄利克雷分布(Latent Dirichlet allocation, LDA)分析消費者評論資料,來更深入地瞭解消費者的體驗,與識別評論背後的消息意圖,並結合服務藍圖(Service Blueprint)服務流程的設計工具,達到企業、消費者價值共創互動的機制。 本文研究的數據集來自中國知名的飲食評論平台:25,670個店家共266,544條線上評論,以及中國外送平台的消費者評論,並將這些評論資料依餐飲服務接觸型態區分成三類,包括:餐桌服務、櫃檯式服務、外送服務,使用隱含狄利克雷分布演算法(Latent Dirichlet allocation, LDA),其為非監督式的機器學習技術,來自動提取顧客在不同餐飲型態下的意見回饋,進行研究與滿意度因子分析。 最終LDA主題演算法萃取出餐飲三種不同型態下,各12項顧客評論中所關心的關鍵主題構面,接著與過去研究比較,發現本研究之方法所歸納出的主題,是可以涵蓋與解釋過去研究的結果,並有其可信度;更重要的是,還找到了原研究中幾乎沒有出現過的新的滿意度因子,包括事前準備資訊、優惠活動、新品、外送天氣、下單操作,這些都豐富了企業決策的數據庫。 將主題分析構面應用服務藍圖(Service Blueprint)與滿意度評分做進一步分析,洞察出商業意涵。在分析中發現不論哪種型態的服務,在負面評論中都是與顧客服務的相關主題(桌邊服務、店員服務、客服)占比最大,由此可以推論,不論與顧客接觸時間長短,人與人間的服務品質是影響顧客情緒產生的重要因素。而在餐桌服務的正面評論中「餐廳內裝」是最高比例會被提及;櫃臺式服務則是「優惠活動」;「外送天氣」最常在外送服務中提到,這幾項主題都會使消費者很有感的產生正面感受,因此企業可以掌握這些資訊做內部管理與外部行銷。 透過分析大量的用戶生成內容來了解消費者的真實聲音,對於競爭激烈的餐飲行業具有重要意義。此次結果資料也可以提供給未來AI客服系統前置訓練的資料建置,來準確歸納消費者的疑問,創造消費者與餐廳價值共創的新服務架構。

並列摘要


In the past, managers often used customer satisfaction surveys to evaluate customer satisfaction and service quality in the catering industry. However, to wait customer filled out questionnaires is a passively method. Therefore, this method can only provide static information collection, and it is difficult to understand the dynamic experience and feedback of customers in the service process. Nowadays, more and more businesses have adopted Service-Dominant Logic (S-D Logic) to replace Goods-Dominant Logic (G-D Logic). This advocates customer-centricity and emphasizes that companies can create value with customers (Lusch Vargo, 2006; Lusch, Vargo, Tanniru, 2010; Vargo Lusch, 2016; Vargo, Maglio, Akaka, 2008). This study, we use data mining methods, Latent Dirichlet Allocation (LDA) and Service Blueprint to analyze the data set, which comes from consumer reviews of well-known food review platforms and delivery platforms in China. This data shows the voices of consumers, to analyze consumers’ experiences, and even identify what’s behind the reviews. In the end, the outcome of LDA shows the new satisfaction factors that does not appear in previous studies, including pre-preparation information, preferential activities, new products, delivery weather, and ordering operations. Understanding the voice of consumers by analyzing a large amount of user online reviews is of great significance to the highly competitive catering industry. The result data can also be provided for construction of the pre-training of the AI customer service system in the future. It can accurately summarize the questions of consumers and create a new service structure that creates value between consumers and restaurants.

參考文獻


英文文獻
1. Andreu, L., Sánchez, I. and Mele, C. (2010), “Value co‐creation among retailers and consumers: new insights into the furniture market”, Journal of Retailing and Consumer Services, Vol. 17 No. 4, pp. 241‐50.
2. Ariyasriwatana, W., Buente, W., Oshiro, M., Streveler, D. (2014). Categorizing health-related cues to action: using yelp reviews of restaurants in Hawaii. New Review of Hypermedia and Multimedia, 20(4), 317-340.
3. Arizon, V. (2010), Service Quality Delivery in the Food and Beverage Industry in the Western Cape., Magister Technologiae: Quality in the Faculty of Engineering, Cape Peninsula University of Technology
4. Back, K. (2012). Impact-range performance analysis and asymmetry analysis for improving quality of Korean food attributes. International Journal of Hospitality Management, 31(2), 535-543.

延伸閱讀