線上評論是廠商了解消費者實際使用經驗的重要資訊來源，而龐大的使用者生成內容（User-generated content, UGC）經常造成廠商資訊超載，大幅提高汲取有用資訊的困難度。為了找出有幫助性評論，近年來，文字探勘與機器學習技術被大量應用於消費者意見內容發掘及預測評論的幫助性。
本研究以台灣地區一家主要的美妝網站評論為資料集，利用集成學習（Ensemble learning）領域中新興的「極限梯度提昇（XGBoost）」技術，建立以評論人氣為依據的預測模型，搜尋對評論人氣有顯著影響的重要字詞，並與傳統線性迴歸技術比較其預測效能。接著透過文字探勘中字詞向量模型的「跳躍式模型（Skip-gram model）」技術找出關聯字詞，並據此偵測出關鍵評論者。同時，在字詞選擇方法中，將「重要單詞選擇法」及「隨機單詞選擇法」的選擇效果進行比較。
研究結果表明，在預測模型中，使用XGBoost技術可以有效提供對評論人氣預測效能更好的字詞。至於在字詞選擇方法中，本研究所提出的「重要單詞結合關聯字詞選擇法」，相較於業者一般以「重要單詞選擇法」或「隨機選擇法」，對人氣評論的預測效果亦顯著較佳。因此，本研究的發現（一）可協助廠商以特定字詞詞組，撰寫產品資訊或評論，吸引較高的評論人氣；（二）可藉此發掘出能有效吸引消費者目光的寫手或評論者，由其撰寫評論，進一步凸顯產品特色以增加消費者的購買意願。

Online reviews are the important information resource for retailers to understand the actual using experience from consumers. However, with the amount of User-generated content (UGC) grows increasingly, retailers would suffer from the problem of information overload to extract the useful information. In order to find out the helpful reviews, the text mining and the machine learning techniques are wide applied to discovering the opinion contents of consumers as well as predicting the helpfulness of reviews in recent years.
This research collected data from one of the main cosmetics and skin care products review website in Taiwan, and utilize the novel technique in the field of ensemble learning called “XGBoost”, building the prediction model which take review popularity as criteria, searching the important terms that have significant effect to review popularity. And compare with the conventional linear regression technique of predictive performance. Then use the “Skip-gram model” technique of word vector modeling in the text mining field to discover the correlatively semantic terms, detecting the key reviewers with both the important terms and the semantic terms. Finally, comparing the performance of the proposed term selected method with the “single term selected method” and “random term selected method”.
The results indicated that XGBoost can further suggest better determinants which have a greater effect on review popularity. In the term selected methods, the proposed method outperforms the other two methods in terms of predicting the review popularity. The finding of this research would help retailers discover the key reviewers who can attract people’s attentions effectively with specific terms group for emphasizing the features of product and promoting the purchase intention of consumers.

Archak, Nikolay, Anindya Ghose, and Panagiotis G. Ipeirotis (2011), “Deriving the Pricing Power of Product Features by Mining Consumer Reviews,” Management Science, 57 (8), 1485–1509.
Cao, Qing, Wenjing Duan, and Qiwei Gan (2011), “Exploring determinants of voting for the ‘helpfulness’ of online user reviews: A text mining approach,” Decision Support Systems, 50 (2), 511–521.
Chen, Jinchuan, Yueguo Chen, Xiaoyong Du, Cuiping Li, Jiaheng Lu, Suyun Zhao, and Xuan Zhou (2013), “Big data challenge: A data management perspective,” Frontiers of Computer Science, 7 (2), 157–164.
Chen, Tianqi and Carlos Guestrin (2016), “XGBoost: A Scalable Tree Boosting System,” in Proceedings of the 22nd ACM SIGKDD International Conference on Knowledge Discovery and Data Mining - KDD ’16, 785–794.
Chevalier, Judith A and Dina Mayzlin (2006), “The Effect of Word of Mouth on Sales: Online Book Reviews,” Journal of Marketing Research, 43 (3), 345–354.
Decker, Reinhold and Michael Trusov (2010), “Estimating aggregate consumer preferences from online product reviews,” International Journal of Research in Marketing, 27 (4), 293–307.
Forman, Chris, Anindya Ghose, and Batia Wiesenfeld (2008), “Examining the relationship between reviews and sales: The role of reviewer identity disclosure in electronic markets,” Information Systems Research, 19 (3), 291–313.
Ghose, Anindya and Panagiotis G. Ipeirotis (2011), “Estimating the helpfulness and economic impact of product reviews: Mining text and reviewer characteristics,” IEEE Transactions on Knowledge and Data Engineering, 23 (10), 1498–1512.
Guyon, Isabelle and André Elisseeff (2003), “An Introduction to Variable and Feature Selection,” Journal of Machine Learning Research (JMLR), 3 (3), 1157–1182.
Krishnamoorthy, Srikumar (2015), “Linguistic features for review helpfulness prediction,” Expert Systems with Applications, 42 (7), 3751–3759.
Li, Jin and Lingjing Zhan (2011), “Online Persuasion: How the Written Word Drives WOM-Evidence from Consumer-Generated Product Reviews,” Journal of Advertising Research, 51 (1), 239–257.
Mudambi, Susan M. and David Schuff (2010), “What makes a helpful online review? A study of customer reviews on amazon.com,” MIS Quarterly: Management Information Systems, 34 (1), 185–200.
Nassirtoussi, Arman Khadjeh, Saeed Aghabozorgi, Teh Ying Wah, and David Chek Ling Ngo (2014), “Text Mining for Market Prediction: A Systematic Review,” Expert Systems with Applications, 41 (16), 7653–7670.
Pan, Yue and Jason Q. Zhang (2011), “Born Unequal: A Study of the Helpfulness of User-Generated Product Reviews,” Journal of Retailing, 87 (4), 598–612.
Park, Do-Hyung, Jumin Lee, and Ingoo Han (2007), “The Effect of On-Line Consumer Reviews on Consumer Purchasing Intention: The Moderating Role of Involvement,” International Journal of Electronic Commerce, 11 (4), 125–148.
Singh, Jyoti Prakash, Seda Irani, Nripendra P. Rana, Yogesh K. Dwivedi, Sunil Saumya, and Pradeep Kumar Roy (2017), “Predicting the ‘helpfulness’ of online consumer reviews,” Journal of Business Research, 70, 346–355.
Tripathy, Abinash, Ankit Agrawal, and Santanu Kumar Rath (2016), “Classification of sentiment reviews using n-gram machine learning approach,” Expert Systems with Applications, 57, 117–126.
Wu, Jianan (2017), “Review popularity and review helpfulness: A model for user review effectiveness,” Decision Support Systems, 97, 92–103.