CAPTCHA = Completely Automated Turing Test To Tell Computers and Humans Apart

The most characteristic properties of CAPTCHA are the following:

  • test instances and their solutions are generated automatically;
  • most humans can pass the test in a very short time;
  • it is hard to write a computer program that has high success over the test in question;
  • the underlying problem must be hard against programs that run for a long time, and against programs that will be developed in the future.

In our opinion looking for reliable human and bot users differentiation we should move to higher levels of human data processing than simple sensory processing. We focus on linguistic competence: our aim is to design a reading CAPTCHA based not only on text recognition but also on text understanding.


The idea of SemCAPTCHA ("Sem" stands for "semantic") is to engage in an authorization process not only sensory processing of a user but linguistic competence as well. A user is presented with a slightly distorted picture of three words. The task is to find one of them which differs from the other two and to point it with a mouse click. The words do not differ substantially as for their graphical properties (like, e.g. length). The difference is of semantic character: one word differs from the other two in its meaning.

Differetiation between human and bot users in CAPTCHA system is based on adequacy of solution as well as on amount of time needed to solve a test instance. Even if a computer program can solve a task at all it takes much more time than in case of a human. In order to strenghten this effect we decided to test if positive semantic priming has a significant impact on time and adequacy of solving this particular kind of lexical tasks by humans. An inspiration comes from cognitive psychology.

An example of SemCAPTCHA task (step-by-step, blue notes are comments)


This research was partially supported by AMU Faculty of Social Sciences
grant No. WSO/133/2006.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License