An introduction to ROC analysis
Pattern Recognition Letters - Special issue: ROC analysis in pattern recognition
Pattern Recognition and Machine Learning (Information Science and Statistics)
Pattern Recognition and Machine Learning (Information Science and Statistics)
A Normalized Levenshtein Distance Metric
IEEE Transactions on Pattern Analysis and Machine Intelligence
An Overview of the Tesseract OCR Engine
ICDAR '07 Proceedings of the Ninth International Conference on Document Analysis and Recognition - Volume 02
NOYB: privacy in online social networks
Proceedings of the first workshop on Online social networks
A low-cost attack on a Microsoft captcha
Proceedings of the 15th ACM conference on Computer and communications security
Textured agreements: re-envisioning electronic consent
Proceedings of the 5th Symposium on Usable Privacy and Security
Machine learning attacks against the Asirra CAPTCHA
Proceedings of the 5th Symposium on Usable Privacy and Security
Adapting the Tesseract open source OCR engine for multilingual OCR
Proceedings of the International Workshop on Multilingual OCR
Americans' attitudes about internet behavioral advertising practices
Proceedings of the 9th annual ACM workshop on Privacy in the electronic society
Proceedings of the 26th Annual Computer Security Applications Conference
Re: CAPTCHAs: understanding CAPTCHA-solving services in an economic context
USENIX Security'10 Proceedings of the 19th USENIX conference on Security
Scramble! your social network data
PETS'11 Proceedings of the 11th international conference on Privacy enhancing technologies
Text-based CAPTCHA strengths and weaknesses
Proceedings of the 18th ACM conference on Computer and communications security
| Hi-index | 0.00 |
This paper presents 'For Human Eyes Only' (FHEO), our Firefox extension that enables one to conveniently post online messages, such as short emails, comments, and tweets in a form that discourages automatic processing of these messages. Similar to CAPTCHA systems, FHEO distorts the text to various extents. We provide a security analysis of its four default distortion profiles as well as a usability analysis that shows how these profiles affect response time and accurate understanding. Our results illustrate the security/usability tradeoffs that arise in the face of adversaries that use current, off-the-shelf optical character recognition technology in order to launch a variety of attacks. Two profiles, in particular, achieve a level of protection that seems to justify their respective usability degradation in many situations. The 'strongest' distortion profile, however, does not seem to provide a large additional security margin against the adversaries we considered.