驗證碼(CAPTCHA),全稱為全自動區分電腦和人類的公開圖靈測試 (Completely Automated Public Turing test to tell Computers and Humans Apart),是當今網際網路中為了對應自動化腳本工具,並防止機器人惡意存取網站資訊,所衍生出的一種工具。有許多提供線上服務的網站,為了確保使用者是真人,均會在存取服務時使用驗證碼進行測試,以維護網站之安全性。常見的應用範例有:銀行網站、學校網站、購票網站或電商網站等。
然而我們發現,現存驗證碼的設計,可能不一定適合應用在所有類型的網站上。在瀏覽過程中,突然顯示出身無關內容的不協調感;以及在驗證碼解題過程中,讓使用者需要花費額外注意力的注意力浪費等問題,都可能是對使用者瀏覽網站的體驗產生負面影響的因素。而這兩點對於電商網站的影響又更加重大,隱含了讓潛在使用者流失的風險。
因此,在這篇論文中,我們提出 YouCaptcha,一種將商品與驗證碼結合的新型驗證碼服務。透過將電商網站本身的商品作為驗證碼題目,解決前述提到的不協調感與注意力浪費之問題。並從注意力、好奇、互動與成就四個方面來探討 YouCaptcha 的設計如何讓使用者對商品產生印象甚至好感。
最後,我們透過使用者研究來驗證我們的假說。在實驗中,我們招募了 56位使用者實地進行測試,結果顯示,雖然在最後的購買率上並沒有產生顯著的影響,但 YouCaptcha 在使用者對商品的記憶以及點擊率的提升都有一定的效果,從實驗的訪談中,多數受測者也認為 YouCaptcha 的設計具有簡單、直覺、美觀等優點,因此我們認為,YouCaptcha 作為針對電商網站設計的新型態驗證
碼,已展現一定程度之效果,並在未來也仍有持續發展並改善的空間。

Completely Automated Public Turing test to tell Computers and Humans Apart, also known as CAPTCHA, is a widely used tool to prevent adversarial bots from accessing website’s resources. Many websites that prioritize security and wish to reduce spams use CAPTCHA to protect their services, such as online transaction services, email services or ecommerce websites.
However, we have observed that the design of current CAPTCHA systems may
not be suitable for all websites. Frictions caused by appearance of unrelated contents, and the waste of attention caused by the additional attention users spend on CAPTCHA, may be two hidden factors that negatively affect the websites, especially for ecommerce websites.
As a result, this paper proposed YouCAPTCHA, a new CAPTCHA system
designed for ecommerce websites. YouCAPTCHA integrates products into
CAPTCHA to resolve the friction and waste of attention problem, and we also examine the design of YouCAPTCHA in four different aspects, in order to make it more profitable for ecommerce websites.
Finally, we evaluate the effectiveness of YouCAPTCHA by conducting an user study. The result showed that, though YouCAPTCHA didn’t have a significant impact on conversion rate, but have a positive effect on both click rate and the recall of products.

摘要 ....................... 1
ABSTRACT .................. 2
目錄 ....................... 3
第一章 前言 ................. 5
第二章 動機 ................. 10
第三章 設計 ................. 12
第四章 假說 ................. 21
第五章 實驗設計 .............. 23
第六章 實驗結果 .............. 30
第七章 討論 ................. 39
第八章 結論 ................. 41
參考資料 .................... 42

[1] Guixin Ye, et al. “Yet Another Text Captcha Solver: A Generative Adversarial Network Based Approach” CCS '18.
[2] Binbin Zhao, et al. “Towards Evaluating the Security of Real-World Deployed Image CAPTCHAs” AISec '18.
[3] “Having a CAPTCHA is Killing Your Conversion Rate”,
https://moz.com/blog/having-a-captcha-is-killing-your-conversion-rate
[4] Leon A. Gatys, at el. “Image Style Transfer Using Convolutional Neural Networks” CVPR '16
[5] Leon A. Gatys, Alexander S. Ecker, Matthias Bethge. “A Neural Algorithm of Artistic Style” 2015.
[6] W Kunst-Wilson, Robert B. Zajonc. “Affective discrimination of stimuli that cannot be recognized” Science, 1980.
[7] J Hansen, M Wänke. “Liking what's familiar: The importance of unconscious familiarity in the mere-exposure effect “ Social cognition, 2009
[8] George Loewenstein, “The psychology of curiosity: A review and
reinterpretation” Psychological Bulletin, pp. 116(1):75-98, 1994.
[9] Dijk E.V., Zeelenberg M. “When curiosity killed regret: Avoiding or seeking the unknown in decision-making under uncertainty” Journal of Experimental Social Psychology, pp. 43, 656-662, 2006.
[10] Hsee, Christopher K. Ruan, Bowen. ”The Pandora Effect: The Power and Peril of Curiosity” Psychological Science, pp. 27(5), 659–666, 2016.
[11] Maximilian Altmeyer, et al. “Gamified Ads: Bridging the Gap Between User Enjoyment and the Effectiveness of Online Ads” SIGCHI 2019.
[12] Aronson, E, Mills, J, “The effect of severity of initiation on liking for a group” The Journal of Abnormal and Social Psychology, pp. 59(2), 177–181, 1957.
[13] Jillian Berman, “The IKEA Effect: Study Finds Consumers Over-Value Products They Build Themselves” 2011.