本研究蒐集士林夜市、逢曱夜市和六合夜市PTT和Facebook的網路口碑資料(Electronic Word of Mouth，簡稱eWOM)，並以文字探勘的技術探討夜市消費者的觀光意象及夜市管理者所傳達的資訊内容，以探究夜市管理者所傳達給消費者的資訊和消費者對於夜市之觀光意象是否有落差。本研究運用數位人文服務Taiwan DH系統建立共詞矩陣，並使用UCINET6繪圖軟體繪製共詞關聯圖，以找出各別夜市在消費者心中的觀光意象和管理者欲傳遞給消費者之資訊内容。本研究結果顯示：(1)消費者對於士林夜市、逢曱夜市和六合夜市之美食意象不同，藉由文字探勘技術可迅速歸納消費者對各大夜市之美食意象；(2)在不同的夜市消費者會感受到不同的觀光意象，例如在士林夜市會帶給消費者氣氛很熱鬧的意象；在逢曱夜市消費者會覺得商品價格平價且流行女裝的產品很多樣化；在六合夜市消費者能夠享有很多商品折扣和優惠活動的資訊；(3)PTT的資料來源反映出消費者的美食意象以熱門之排隊美食為主，而Facebook粉絲專頁之資料來源反映出管理者傾向於發布最新商品資訊吸引更多顧客來夜市消費，二者間有明顯差異。

In this study, the electronic word-of-mouth (eWOM) of Shilin, Fengjia, and Liuhe Night Markets are collected from the PTT bulletin board system and Facebook. Accordingly, the consumer’s tourism image of night markets and the information the night market managers convey are analyzed by text mining methods. A digital humanities (DH) service system, called Taiwan DH, is used to analyze co-words and a graphics software, called UCINET6, is employed to draw the association graph by the co-word analysis. On the basis of the association graph, one can identify consumers’ tourism images of the night markets and the information that night market managers convey. Furthermore, a comparative study is conducted to find the gap between consumers’ tourism image and the information the night market managers convey, which could provide a reference for night market managers when setting strategies for sustainable development.