以神秘客調查與社群網站分析探勘消費者行為
No Thumbnail Available
Date
2022
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
隨著社群網路的發展與普及，越來越多的消費者於網路表達自己的想法或意見，也提供更多樣的消費資訊來源。文字探勘為提供研究者整合網路資料，擷取信息進行分析，探勘消費者行為的方法，並可協助企業，調整行銷策略或戰術。而神秘客分析為行銷領域長久以來，廣為運用於分析消費者行為之工具。雖然社群網路與神秘客分析各自有其價值，但少有學者同時以兩種方法分析同一筆資料，並比較結果之差異。因此，本研究擬探勘社群網站，擷取與產品或服務相關貼文之後，透過隱含狄利克雷分佈（Latent Dirichlet Allocation，LDA）擷取主題，以群落分析將主題分群之後，歸入理論模型構面，經專家確認後，以偏最小平方法結構方程模型Partial Least Square Structure Equation Model (PLS-SEM) 驗證路徑。本研究亦同步分析神秘客報告，以相同之文字探勘、服務品質理論模型與結構方程模型驗證路徑之顯著與否。本研究服務品質的探討是以服務提供者端出發，探討在服務提供的過程中可能發生影響服務品質的因素，其影響使用者所知感受與需求，本研究從使用者的角度出發，研究服務形成的過程中，使用者在服務品質模型之顯著性，研究使服務品質模型更加完善。本研究以台灣行動裝置門市與寵物連鎖店之服務為研究對象，以2021年之社交網絡(Dcard.tw)貼文與神秘客分析報告進行分析，比較二分析結果之異同，依據分析結果除有形性及保證性之顯著性有差異外，在可靠性、回應性及同理性上均為顯著影響服務品質之關鍵要素，本結果可作為企業規畫服務、訂定推廣決策參考之用。
In the age of social networks, more and more consumers are expressing their opinions, thoughts, and ideas on the Internet. Text mining offers a method for researchers to retrieve network data and extract information for analysis, explore consumer behavior, and assist firms in defininig or adjusting marketing strategies or tactics. Mystery shopper analysis has long been a tool for ana-lyzing consumer behaviors. However, few scholars compare the differences between the two methods.Therefore, this study intends to mine social networking sites after ex-tracting posts related to products or services, extracting topics using the Latent Dirichlet Allocation (LDA), clustering the topcis into groups using cluster analysis, and classifying them into topcis. After confirming the topcis by ex-perts , and the path was verified using Partial Least Square Structural Equation Modeling (PLS-SEM). This research also analyzes the mystery shoppers report simultaneously based on the same theoretical model. The PLS-SEM is adopted again to confirm the path model.This study takes the services of mobile device stores and pet chains in Taiwan as the research object, analyzes the social network (dcard.tw) posts and mystery shopper analysis reports in 2021, and compares the similarities and differences between the two analysis results. According to the analytic results, except the significanct discrepencies in tangibility and assurance, reliability, responsiveness and homogeneity are the key factors that significantly affect service quality. The results canbe used as a reference for enterprises to plan services and make promotion decisions.
In the age of social networks, more and more consumers are expressing their opinions, thoughts, and ideas on the Internet. Text mining offers a method for researchers to retrieve network data and extract information for analysis, explore consumer behavior, and assist firms in defininig or adjusting marketing strategies or tactics. Mystery shopper analysis has long been a tool for ana-lyzing consumer behaviors. However, few scholars compare the differences between the two methods.Therefore, this study intends to mine social networking sites after ex-tracting posts related to products or services, extracting topics using the Latent Dirichlet Allocation (LDA), clustering the topcis into groups using cluster analysis, and classifying them into topcis. After confirming the topcis by ex-perts , and the path was verified using Partial Least Square Structural Equation Modeling (PLS-SEM). This research also analyzes the mystery shoppers report simultaneously based on the same theoretical model. The PLS-SEM is adopted again to confirm the path model.This study takes the services of mobile device stores and pet chains in Taiwan as the research object, analyzes the social network (dcard.tw) posts and mystery shopper analysis reports in 2021, and compares the similarities and differences between the two analysis results. According to the analytic results, except the significanct discrepencies in tangibility and assurance, reliability, responsiveness and homogeneity are the key factors that significantly affect service quality. The results canbe used as a reference for enterprises to plan services and make promotion decisions.
Description
Keywords
神秘客, 文字探勘, 社交媒體, 主題建模, 偏最小平方結構方程模型, Mystery Shopper, Text Mining, Social Media, Topic Modeling, Partial Least Square Structural Equation Modeling (PLS-SEM)