運用文字探勘挖掘TripAdvisor旅館業評論進行顧客滿意度的研究－以臺灣、中國為例

Using Text Mining to Mine TripAdvisor Hotel Industry Reviews Customer Satisfaction Studies in Taiwan and China

張美淑(Mei-Shu Chang) ； 丁冰和(Ping-Ho Ting)

《休閒產業與健康促進學術研討會》 2018 (2018/06) Pp. 43-60

摘要

因網路興起行動載具便利、旅遊資訊透過平台大爆發，資訊搜尋變得十分容易。現今旅客可以透過網路直接發佈對飯店經驗的評論，線上評論就變成影響消費者選擇飯店的關鍵因素，而大量旅客內心深層、自然呈現的評價與評論，也是洞察商機搶得先鋒的訊息。因此，本研究嘗試從TripAdvisor線上評論中利用意見挖掘(opinion mining)及內容分析 (content analysis)抓取台灣、中國星級旅館飯店業評論，進行分析與討論。TripAdvisor是全球最大旅遊網站，有5億則不偏頗的旅客評論。本研究從平台中抓取消費者對兩地星級旅館業所留下的各300則評論，並將服務滿意度構面分成「人的服務」、「消費物件」及「空間環境」 三大構面，包括員工接待服務態度、交通環境、空間氛圍、房間大小和浴室、床、等設施，將計算相關詞彙出現頻率及情感分析，利用內容分析及文字雲分析(word cloud)找出優先改善次序、 隱藏問題及台灣、中國消費者習性，提供給業者建議和參考，以利做好顧客關係管理，讓兩地業者的服務更貼近消費者的需求，制定重要的商業決策。有別於過去很多顧客滿意度相關研究採用問卷或訪談的方式，未能反映消費者最深層的記憶印象，且是在有限制時間下所做的反應，所得問卷或訪談內容，未必是消費者的真實感受。本研究嘗試從線上評論並採用意見挖掘(opinion mining)及內容分析(content analysis)之技術，期盼 提出更深層及符合業界所分析。

關鍵字

線上評論 ； 意見挖掘 ； 內容分析 ； 顧客滿意度

並列摘要

Due to the Internet's emergence of convenient mobility and tourism information through the platform, information search has become very easy. Visitors can publish reviews of hotel experiences directly via the Internet. Online reviews have become a key factor influencing consumers' choice of hotels. The evaluation and commentary of a large number of passengers' innermost feelings and natural presentations is also a message to gain insight into business opportunities. Therefore, this study attempts to use the opinion mining and content analysis from the online reviews of TripAdvisor to capture the reviews of the Taiwan and Chinese star-rated hotels and restaurants industry for analysis and discussion. TripAdvisor is the world's largest travel website. There are 500 million unbiased traveler's comments. This study captures 300 reviews of consumer's comments of the star-rated hotel industry in the two places from the platform and divides the service satisfaction aspect into three major components: ＂person＂, ＂consumer item＂ and ＂space and environment＂, Including employee reception service attitude, traffic environment, space atmosphere, room size and bathroom, bed, and other facilities, the frequency and sentiment analysis of relevant vocabulary will be calculated. Use content analysis and word cloud to identify priorities for improvement, hidden issues, and consumer behaviors in Taiwan and China. To provide suggestions and references for operators to facilitate the management of customer relationships, so that the services of the two businesses can be closer to the needs of consumers and make important business decisions. Different from the past, many researches on customer satisfaction related to questionnaires or interviews. Failure to reflect the deepest memory impressions of consumers is a response that occurs within a limited period of time. The resulting questionnaires or interviews may not necessarily be the true feelings of consumers. This study attempts to comment online and adopt techniques of opinion mining and content analysis. It hopes to provide deeper insights and analysis in line with the industry.

並列關鍵字

Online reviews ； opinion mining ； content analysis ； customer satisfaction