Human vs. machine: Analyzing the robustness of advertisement based CAPTCHAs

Abstract

A program that can ensure whether the response is generated by a human, not from a machine, is called CAPTCHA. They have become a ubiquitous defense to protect web services from bot programs. At present most of the visual CAPTCHAs are cracked by programs; as a result they become more complex which makes both human and machine finds it difficult to crack. NLPCAPTCHAs, an acronym for Natural Language Processing CAPTCHA, are introduced to provide both security and revenue to websites owners through advertisements. They are highly intelligent CAPTCHAs that are difficult for a machine to understand whereas it's a better user experience when compared to traditional CAPTCHA. In this paper we have introduced a method that is able to analyze and recognize the challenges from three different types of NLPCAPTCHAs. © Springer International Publishing Switzerland 2014.