Proposal and Evaluation of Multi-model CAPTCHA with Typoglycemia

We propose a Multi-model CAPTCHA that is composed of one sentence with typo and multiple pictures. This utilizes human's three abilities (inferring the meaning typo, understanding pictures, linking corresponding picture), it based on hypothesis, "the longer time could be for machine to link picture (= shorter linking, higher probability human)." found findings from our experiment: (1) Examinees were able even there some pictures similar composition, (2) Typoglycemia likely appeared constructed by short familiar words, (3) The was not exponentially increased length getting long It difficult users in this situation).