An investigation of the usability of image-based CAPTCHAs using PROMETHEE-GAIA method

Abstract

Today, it is difficult to find an adequate Web site with a registration form that is not protected with some automated human proof test. One of the oldest concepts in Artificial Intelligence as a security mechanism based on the Turing Test is CAPTCHA (Completely Automated Public Turing test to tell Computers and Humans Apart). This test was designed to make a difference between the real users and bots and provide security against malicious attacks. The PROMETHEE-GAIA method was employed for ranking different image-based CAPTCHAs according to their usability in this paper. The aim of this study is a comparative analysis of seven image-based CAPTCHAs based on three different criteria: time to find a solution, a number of attempts, and task difficulty. The weights of the considered criteria were calculated objectively by the entropy method, and for the subjective weights, Analytical Hierarchy Process (AHP) was used. A defined research model was applied through four phases. The survey included 320 randomly selected Internet users and experts in computer science who were familiar with CAPTCHA tests. The proposed model suggested which CAPTCHA offered better human accuracy and lower machine attack rates compared to the existing approaches. The obtained results were very helpful to the web administrators because they indicated that this approach could provide useful insights for the decision-makers about the appropriate mechanisms to protect users against cyber-criminal activities and Internet threats.