Earticle

초록

영어

We propose statistical methods for evaluating the efficiency of CAPTCHA. Most people unfairly assumed that machines are not capable at reading precisely. This fact leads to the invention of CAPTCHA, a distorted word or short phase, which is designed to thwart computers and separate human from machines. However, advances in image recognition technologies mean that machines are constantly getting better at recognizing CAPTCHA. This forces CAPTCHA designers to design even more difficult CAPTCHAs to prevent their systems from being gamed by malicious bots. However, this arm race has an unintended side effect on the common users. Many CAPTCHAs are now so hard that many people are unable to read them. This obviously conflicts with the original purpose that CAPTCHA was invented in the first place. Our analysis shows that some CAPTCHAs are more users friendly. In particular, Yahoo-style CAPTCHA is the most friendliness. This suggests that a good CAPTCHA could be a simple text with some distortion that prevents machines from correctly segmenting characters.

목차

Abstract
 1. Introduction
 2. Literature Review
 3. Experimental Design
  3.1 Sample Size
  3.2 OCR Engine
  3.3 CAPTCHA
  3.4 Website Challenge Preparation
  3.5 CAPTCHA Efficiency Evaluation
 4. Results
 5. Conclusion
 References

키워드

저자

• Passzarkorn Youthasoontorn [ Department of Computer Engineering, Chulalongkorn University, Bangkok, Thailand ]
• Akarin Phaibulpanich [ Department of Statistics, Chulalongkorn University, Bangkok, Thailand ]
• Krerk Piromsopa [ Department of Computer Engineering, Chulalongkorn University, Bangkok, Thailand ] Corresponding Author

참고문헌

발행기관

간행물

표지보기 관심저널 등록

• 간행물명

  JITAM [Journal of Information Technology Applications and Management]

• 간기

  격월간

• pISSN

  1598-6284

• eISSN

  2508-1209

• 수록기간

  1999~2026

• 십진분류

  KDC 005 DDC 005